# Approvals and Policy
Source: https://docs.decisional.com/agents/approvals
Control when agents pause, ask for review, bypass approvals, or block actions
## Overview
Approvals are Decisional's human-in-the-loop control system. They let an agent prepare work, show the relevant context, and pause before a sensitive action proceeds.
Approvals can appear during automated runs, Chat Mode write actions, Run Mode flows, and workflow gates. They are especially useful when an agent is about to send information, update a system, make a decision with business impact, or continue past a manual review point.
## Where approvals appear
The agent pauses before an action that writes, sends, updates, creates, or deletes information.
A workflow can pause at an explicit review step before moving to the next node.
If an agent wants to take a write action during chat, it can ask for approval before executing.
Stakeholders can review context and approve or reject without needing the full builder interface.
## What an approval contains
An approval should give the reviewer enough context to make a decision without digging through logs.
Approvals typically include:
* The action the agent wants to take
* The relevant run, node, or chat context
* Inputs and generated outputs
* Files or records involved in the decision
* Approve and reject controls
* Any notes the reviewer adds before continuing
When an approval is accepted, the run continues. When it is rejected, the agent stops, routes to the configured rejection path, or asks for a correction depending on the workflow.
See how Dex asks for approval in Slack before creating a Linear issue.
## Approval layers
Approval behavior comes from multiple layers working together.
| Layer | What it controls | Typical options |
| ----------------------------------- | ---------------------------------------------------------- | ------------------------------------------- |
| **Workspace default approval mode** | The account default used when a runner is set to Default | Ask approvals or Bypass approvals |
| **Runner approval picker** | The approval mode for the current chat or run prompt box | Default, Ask approvals, or Bypass approvals |
| **Tool approval policy** | How toolkit read/write tools and per-tool overrides behave | Always allow, Needs approval, or Blocked |
| **Workflow gates** | Manual review points designed into the workflow | Approver list, quorum, approve, reject |
Use these layers together. For example, a low-risk reporting agent might bypass eligible approval prompts, while a finance agent can keep write tools on **Needs approval** and add gate nodes for final review.
## Workspace default and runner picker
The workspace default is the fallback approval mode. It is used only when the runner is set to **Default**.
The prompt-box approval picker controls the runner you are about to use. The **Default** option shows the effective workspace default in parentheses. If the picker says **Default (Bypass approvals)**, choosing Default means eligible approval prompts will be bypassed.
For live runs, choose **Ask approvals** in the runner picker when you want the agent to pause before eligible sensitive actions. If the live runner is set to **Default** and the effective default is bypass, the run continues without approval prompts.
Test runs are the best place to keep Ask approvals on while you review behavior. Live runs only pause for approval when the effective runner setting is Ask approvals, when tool policy requires approval, or when the workflow reaches an explicit gate.
## Ask approvals
**Ask approvals** means the agent pauses before eligible sensitive actions. A reviewer sees what the agent wants to do and chooses whether to approve or reject.
Use Ask approvals when:
* The agent is new or still being tuned
* Outputs leave your company or affect customers
* The workflow changes records in important systems
* A human needs accountability for the final decision
* You want an audit trail of manual sign-off
## Bypass approvals
**Bypass approvals** means eligible approval prompts are approved automatically so the run can continue without waiting for a human click.
Bypass approvals are useful for stable, low-risk workflows where the agent's behavior is well understood. Bypass does not turn blocked tools into allowed tools. A blocked action remains blocked, and tool policy should still be used for actions that must never run.
Use bypass carefully:
* Start new agents with Ask approvals until the workflow is predictable
* Keep destructive or externally visible actions behind tool policy or explicit review
* Review run history after enabling bypass
* Prefer bypass for repeatable internal tasks with clear inputs and low blast radius
## Settings and workspace policy
The Settings page contains the workspace approval default that runners use when they are set to **Default**.
From Settings, admins can manage:
* The default approval mode used when a runner has no override
* Toolkit defaults for read tools and write tools
* Per-tool overrides for actions that should always be allowed, require approval, or be blocked
* Team access and workspace configuration that affects who can approve or operate agents
The effective behavior is the combination of the workspace default, the runner picker, tool policy, and any workflow gates.
Bypass approvals should be treated as an operating policy, not a shortcut around governance. Keep blocked tools blocked, and keep human review on workflows where the business risk requires it.
## Recommended rollout
Keep new agents in Ask approvals while you test the workflow, review outputs, and tune instructions.
In the toolkit tools view, allow safe read actions, require approval for sensitive write actions, and block tools the workspace should not use.
Put explicit workflow gates before final send, publish, payment, record update, or escalation steps.
Once a workflow is stable, enable Bypass approvals for agents where waiting for human review is no longer necessary.
Review run history, failures, and approval outcomes to keep the policy aligned with the workflow's risk.
## Related docs
See how approvals behave across Chat Mode, Operator Mode, and Run Mode.
Learn where action nodes and gate nodes fit inside a workflow.
# Inbox
Source: https://docs.decisional.com/agents/inbox
Give an agent a dedicated email address and review its conversations
## Overview
Inbox gives an agent a dedicated Decisional-managed email address. It is useful when the agent should send from its own identity or receive email threads that can be reviewed from Operator Mode, instead of using a person's Gmail or Outlook mailbox.
## Create an inbox
Open **Inbox**, enter a unique name for the address, and wait for Decisional to confirm that it is available. The product adds the displayed email domain.
Select **Create inbox**. The address is attached to this agent and can be copied from the Inbox header.
Add the AgentMail tool when the workflow needs to send from this inbox. An inbox is Decisional-managed and does not require Gmail or Outlook OAuth.
## Review email threads
After messages arrive, Inbox groups them into threads and shows inbound and outbound messages together. Filter the list by **All**, **Received**, **Unread**, or **Sent**. Open a thread to read the full history, inspect attachments, and see message status. New and updated threads appear in real time, and you can also refresh the list manually.
## Manage inbox access
Inbox settings let you enable or disable the address, set its sender policy to **Anyone** or an allowed list, and delete the inbox. Individual threads can also be deleted from the review surface.
Deleting an inbox or email thread is permanent. Confirm that the history is no longer needed before deleting it.
## Related docs
Learn how Inbox fits into the complete agent workspace.
Learn how tools and credentials are connected to an agent.
# Instructions
Source: https://docs.decisional.com/agents/instructions
Understand how instructions connect integrations, tools, credentials, and agent versions
## What Are Agent Instructions?
Instructions are the operating guide for an agent. They explain the goal, the order of work, the rules to follow, and what a successful result looks like.
Instructions also connect the written process to the exact tools the agent can run. An instruction can say “enrich this person,” but the tool declaration tells Decisional to use a specific Apollo operation with a specific credential.
In this example:
* The instruction body explains the daily outreach process in plain English
* The chips above the instructions show the integrations used by the agent
* **Apollo** and **Exa** use Decisional's platform credentials
* **Gmail**, **Google Sheets**, and **Slackbot** use connected accounts
* The right panel shows the exact Apollo tool, **Enrich person with Apollo**, and the credential assigned to it
* The lock beside the agent name shows that the Live version is read-only
## Integrations
An **integration** is the connection layer for an app or service, such as Apollo, Gmail, Google Sheets, or Slack.
The integration chips above the instructions give you a quick connection summary:
| Label | What it means |
| ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Platform** | The selected tools use a Decisional-managed credential. No provider account is needed for those tools. Usage may be metered through workspace credits. |
| **Connected** or a check mark | The tools are mapped to an account connected by you or your workspace. |
| **Connect** | At least one required tool does not have a usable credential yet. |
Click an integration chip to open **Connect Integrations** in the right-hand panel over the chat. The panel lists the credentials, tools, and triggers attached to that integration without taking you away from the instructions. With the panel open, hover over an integration chip or integration row to highlight its matching location on the other side.
Adding an integration does not give the agent every action in that app. The agent receives only the tools declared for its instructions and workflow.
## Tools Declared in Instructions
A **tool** is one specific operation inside an integration. Gmail is an integration; **Create email draft** and **Send email** are two different Gmail tools. Apollo is an integration; **Enrich person with Apollo** is one Apollo tool.
Dex declares the exact tool where it is needed in the instructions. The editor displays that declaration as an inline tool chip.
Hover over an inline tool chip to highlight the same tool in the right-hand connection panel. Hover over a tool in the connection panel to highlight the exact place it is declared in the instructions. Both locations receive a visible outline, making it easy to follow the tool from the written step to its credential mapping.
Click an inline tool chip to open its integration in the right-hand panel over the chat. The matching tool is shown with the credential it will use, such as a specific Gmail account or the Decisional **Platform** credential.
### Find Where a Tool Is Used
Click an integration chip above the instructions. The connection panel opens on the right while the instructions remain visible.
Open the integration to see every tool declared for this agent and the credential selected for each one.
Hover over a tool in the right-hand panel. Decisional highlights the matching inline declaration in the instructions, showing exactly where that tool is used.
Read the credential selector directly below the tool. It shows whether that instruction uses a platform credential, a connected account, a simulated credential, or no credential.
For example, an outreach agent might use these declarations:
| Instruction step | Declared tool |
| --------------------------- | ----------------------------- |
| Enrich the selected contact | **Enrich person with Apollo** |
| Create a personalized draft | **Create Gmail draft** |
| Mark the row as drafted | **Update Google Sheets row** |
| Post the completion summary | **Send Slackbot message** |
Behind the editor, each tool has a stable tool name such as `APOLLO_PEOPLE_ENRICHMENT`. Dex resolves that name into an inline tool chip and a tool instance for the agent. You normally do not need to type tool names yourself—tell Dex which action the instructions should perform.
Be precise about the action. “Use Gmail” is ambiguous. “Create a Gmail draft and do not send it” tells Dex which tool and safety behavior the workflow needs.
## Credentials Mapped to Tools
A **credential** authorizes a tool to use a particular account. Credentials are mapped to tools, not pasted into the written instructions.
Each tool declaration has its own mapping:
| Tool | Possible credential mapping |
| ------------------------- | ----------------------------------- |
| Enrich person with Apollo | Decisional **Platform** credential |
| Create Gmail draft | `ravi@company.com` Gmail connection |
| Read leads sheet | Company Google Sheets connection |
| Send completion message | Workspace Slackbot connection |
This per-tool mapping matters when an integration has more than one connected account. Two Gmail tool declarations can use different mailboxes, even when both belong to the Gmail integration.
When Dex adds a tool, Decisional tries to bind it in this order:
1. A compatible user or workspace credential that is already connected
2. A supported platform credential when the tool is platform-provided
3. No credential, which leaves the tool waiting for you to connect or select an account
Use the dropdown beneath a tool in the connection panel to choose a different compatible account or switch between your own account and a platform credential when supported.
An integration can look available while one of its tools still needs a credential. Review every tool in the connection panel before testing the agent.
## Multiple Credentials for the Same Tool
An agent can use more than one credential from the same integration. Each tool declaration has its own credential mapping, even when two declarations use the same underlying tool.
For example, one agent could use Gmail like this:
| Instruction step | Tool | Credential |
| ---------------------------------- | --------------------- | --------------------------- |
| Read new support requests | Search Gmail messages | `support@company.com` |
| Check conversations with prospects | Search Gmail messages | `sales@company.com` |
| Create an approved reply | Create Gmail draft | `account-owner@company.com` |
The first two steps use the same Gmail search tool, but they are separate declarations and can read from different mailboxes. Ask Dex to declare the tool separately for each purpose, then select the intended account beneath each tool in the connection panel.
The same pattern works for other integrations—for example, reading from two Google Sheets accounts, posting to different Slack workspaces, or using a platform credential for one Apollo read while another Apollo tool uses your connected account.
### Swap a Credential Without Rewriting the Workflow
A workflow node declares **which tool operation to call**. The credential mapping decides **which account performs that operation**. Because those are separate, you can change a tool's account without rewriting the instruction or changing the node's logic.
If the agent is Live, Tested, or otherwise locked, click the lock beside its name to create an editable Draft.
Click the integration or inline tool chip to open the right-hand connection panel.
Hover over each tool in the panel to highlight where that instance is used in the instructions.
Use the dropdown below that tool to select another compatible connected account or a platform credential when available.
Run a small test to confirm the new account can access the expected data and perform the action. Build and publish the Draft when it is ready.
Changing the mapping keeps the same tool, instructions, and workflow node. It can still change the result because accounts may contain different data, use a different sender identity, or have different permissions. Always verify the new mapping before publishing, especially for email, CRM, file, or messaging actions.
Credentials are not stored inside the workflow node. The node keeps its tool declaration, while Decisional resolves the selected credential securely when the tool runs.
## Locking and Unlocking an Agent
The lock protects a stable agent configuration from accidental edits.
An agent is read-only when it is in a built, tested, Live, or Paused state. It is also temporarily locked while Decisional is building, patching, or testing it. While locked, you can still read the instructions and click tool chips to inspect their mappings, but you cannot edit the instructions, add integrations, or change credentials.
### Unlock an Agent
1. Open the **Instructions** tab in Operator Mode.
2. Click the lock icon beside the agent name.
3. The agent returns to **Draft**, where you can edit its instructions and connections.
4. Build and test the updated draft.
5. Publish it when the new version is ready.
Unlocking a Live or Paused agent creates a draft from the deployed version. The previous version remains available in version history, but the agent itself returns to Draft while you edit. Rebuild, test, and publish the draft before expecting scheduled or triggered production work to use the changes.
A temporary lock shown during **Building**, **Patching**, or **Testing** cannot be removed. Wait for the current operation to finish.
## How the Pieces Fit Together
| Concept | Answers this question |
| ---------------------- | ------------------------------------------------------------------------------ |
| **Instructions** | What should the agent do, and what rules should it follow? |
| **Integration** | Which app or service provides the capability? |
| **Tool declaration** | Which exact operation can the agent perform at this point in the instructions? |
| **Credential mapping** | Which authorized account will that tool use? |
| **Lock** | Is this version protected from edits or open as a Draft? |
Before publishing, read the instructions from top to bottom, open every integration, confirm each tool has the intended credential, and run a small test.
## Related Documentation
Learn how Apollo, Exa, Agent Mail, and Linq SMS platform credentials work.
Learn how accounts, tool bindings, and workspace access are managed.
Learn how drafts, versions, testing, and publishing work together.
See how instructions become an executable workflow graph.
# Modes
Source: https://docs.decisional.com/agents/modes
Understand Chat Mode, Operator Mode, and Run Mode for agents
## Agent modes
Agents can be used through different surfaces depending on the job:
* **Chat Mode** is for conversation, one-off work, explanations, and follow-up questions.
* **Operator Mode** is for changing how the agent works.
* **Run Mode** is for public run links where someone can run or review the agent without the builder surface.
Each agent owns instructions, a workflow, tools, sources, triggers, runs, and approval policy. The mode picker controls which surface you use for that agent.
## Agent operating modes
An agent can be used in different modes depending on who is using it and what they need to do.
| Mode | Use it for | Who should use it |
| ----------------- | ----------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------- |
| **Chat Mode** | Ask questions, run one-off tasks, inspect context, and iterate conversationally | Anyone collaborating with the agent |
| **Operator Mode** | Configure instructions, workflow, tools, triggers, skills, sources, runs, and approval behavior | Owners and builders of the agent |
| **Run Mode** | Share a public link where people can view and run a public agent | External stakeholders or internal users who do not need the builder surface |
## Chat Mode
Chat Mode is the everyday conversational surface for an agent. It is best when you want to ask the agent to explain a result, process a small request, answer questions over its sources, or continue an existing thread.
Chat Mode keeps configuration changes controlled. If a request requires editing integrations, workflow structure, triggers, or other operator-level settings, the agent will route you to Operator Mode.
## Operator Mode
Operator Mode is the builder and control room for an agent. Use it when you need to change how the agent works, not just ask it to do work.
Operator Mode includes:
* Agent instructions and build context
* Workflow viewer and node configuration
* Connected tools and credentials
* Triggers and schedules
* Skills and sources
* Runs, logs, failures, and pending reviews
* Agent-level approval settings
Use Operator Mode for setup, debugging, and policy changes.
## Run Mode
Run Mode is the shareable run experience for an agent. When an agent is made public, anyone with the link can view and run that agent from the focused public surface.
Run Mode is useful for:
* Intake workflows
* Customer-facing request flows
* Vendor or partner review
* Internal forms where most users should not edit the agent
* Approval pages where a stakeholder only needs the context and decision controls
Public access does not remove the agent's approval policy. Tool policies, approval requirements, and blocked actions still apply.
## Choosing a mode
Use Chat Mode.
Use Operator Mode.
Use Run Mode after making the agent public.
## Related docs
Learn how approval modes, bypass approvals, and global policy work.
Build and manage agents end to end.
# Outputs
Source: https://docs.decisional.com/agents/outputs
Find, preview, and download files created by agent runs
## Overview
Outputs are files and artifacts created by agent runs. Examples include a generated PDF, spreadsheet, presentation, image, or exported report.
## Browse outputs
The **Folders** view groups files by their virtual path, which commonly includes the run that created them. Select a folder to see its files, size, run, and creation time. Use **List** to scan all outputs in one table.
## Preview and download files
Click a file to preview it. Spreadsheet files open in a workspace tab, while supported documents open in the document preview. Use the download button when you need a local copy. The run link helps you trace an artifact back to the execution that produced it.
An output is a deliverable, not memory. Saving a file does not replace State Store when the next run needs a reliable de-duplication marker, counter, cursor, or status value.
## Related docs
Learn where agents keep small pieces of durable operational memory.
Trace an output back to the run that created it.
# Sources
Source: https://docs.decisional.com/agents/sources
Give your agents context with documents, files, and web links
## What Are Sources?
Sources are documents and files that give your agent the context it needs to do its job. Think of them as the reference materials you'd hand a new team member — price lists, SOPs, contracts, policies, or any document your agent needs to reason over.
When you upload a source, Decisional indexes it using **Retrieval-Augmented Generation (RAG)**. Your agent can then search and retrieve relevant information from your documents at runtime.
## Supported File Types
Contracts, invoices, forms, reports, SOPs (up to 50MB, OCR supported)
Policies, templates, procedures (Word, PowerPoint)
Receipts, forms, diagrams (PNG, JPG, TIFF up to 20MB with OCR)
Code, configs, raw text, JSON, XML
## Adding Sources
In your agent page, click **Sources** in the left sidebar.
Click **Add Source** and choose:
* **Upload File** — drag and drop files (up to 50MB each)
* **Add Link** — paste a URL and Decisional will fetch and index the content
Decisional indexes your source (usually 10 seconds to 3 minutes depending on size). Once indexed, your agent can use it.
Tell your agent when to use specific sources:
```
Reference the "Q4 Price List.pdf" for current pricing.
Use the "Vendor Catalog.xlsx" to validate vendor names.
```
## How RAG Works
When your agent runs, it doesn't read every document end to end. Instead, it uses RAG to find the most relevant sections:
1. **Indexing** — your documents are chunked and semantically indexed when uploaded
2. **Retrieval** — at runtime, the agent searches for sections relevant to the current task
3. **Generation** — the agent uses the retrieved context to reason and produce output
4. **Citation** — the agent cites the specific pages and sections it referenced
## Bounding Box Citations
For PDFs, Decisional shows you the **exact location** the agent read — including the page number, paragraph, and a highlighted bounding box around the text. This makes it easy to verify your agent's reasoning.
## Tips for Better Source Quality
* **Use text-based PDFs** — scanned images work (via OCR) but text-based PDFs are faster and more accurate
* **Remove irrelevant pages** — cover pages, blank pages, and appendices add noise
* **Use clear formatting** — documents with headers and sections are easier for agents to navigate
* **Keep documents focused** — one topic per file works better than large catch-all documents
* **Use descriptive filenames** — "Q4 2025 Price List" is better than "pricelist\_final\_v3.xlsx"
* **10-20 sources is the sweet spot** — too many sources can slow retrieval
Start with fewer, high-quality sources and add more as needed. Quality beats quantity.
# State Store
Source: https://docs.decisional.com/agents/state-store
Inspect the durable state an agent keeps between runs
## Overview
The State Store is the agent's durable memory for small pieces of structured data that must survive between runs. Workflows are otherwise stateless: if a later run needs to remember what an earlier run already processed, the workflow must save that information here.
Good uses include:
* de-duplication markers, such as IDs that were already processed
* cursors, counters, flags, and last-run timestamps
* short-lived locks or values that should expire automatically
* progress that another node or future run needs to continue
State is stored as JSON under a **namespace** and **key**. Namespaces keep unrelated data separate. Each entry also has a version, update time, and optional expiry.
## Inspect saved state
Open **State Store** from Agent Manager to inspect what the agent currently remembers. You can search keys and values, filter by namespace, refresh the list, and expand a row to read the full JSON value.
This screen is an inspection surface. The workflow's state tools create, update, or delete entries while it runs.
Use State Store for small operational memory, not reports or large files. Put documents, spreadsheets, images, and other deliverables in Outputs instead.
## Related docs
Learn where files and other run deliverables are stored.
Inspect the runs that read or update agent state.
# Versioning
Source: https://docs.decisional.com/agents/versioning
Switch between versions, edit drafts safely, and restore previous configurations
## Why Versioning?
When your agent is live and running, you don't want to break it while making changes. Agent versioning lets you edit a draft copy of your agent without affecting the live version. When you're happy with the changes, publish the new version.
## How It Works
Every time you publish an agent, Decisional saves a snapshot of its configuration — instructions, workflow, integrations, and triggers. This becomes a **version**.
* **Live version** — the version currently running in production
* **Draft version** — your working copy where you make changes safely
Changes to a live agent are always made in a draft first. Nothing goes live until you explicitly publish.
## Editing a Live Agent
When you want to update a live agent:
Navigate to the agent you want to edit. You'll see its current live version.
Any edits you make — updating instructions, changing integrations, modifying the workflow — are saved as a draft. Your live agent keeps running the previous version untouched.
Run a test on the draft version to make sure everything works as expected.
When you're satisfied, hit **Build** and then **Publish** to push the new version live. Your agent immediately starts using the updated configuration.
## Switching Between Versions
Use the **version selector** on the agent page to browse your version history. You can:
* **View any past version** — see exactly what instructions, workflow, and integrations were active at that point
* **Restore a previous version** — roll back to an earlier configuration if something goes wrong
* **Compare versions** — understand what changed between versions
If a new version causes issues, you can quickly restore the previous version to get your agent back to a working state.
## What Gets Versioned
Each version captures a full snapshot of your agent:
| Component | Versioned? |
| -------------------------- | ---------- |
| Instructions | Yes |
| Workflow (nodes & edges) | Yes |
| Integration connections | Yes |
| Triggers & schedules | Yes |
| Sources and attached files | Yes |
## Best Practices
* **Test before publishing** — always run a test on your draft before pushing it live
* **Make incremental changes** — small, focused updates are easier to debug than large rewrites
* **Use versions as checkpoints** — before a major change, make sure your current working version is published so you can roll back
* **Copy your Agent ID** — you can copy your agent ID from the page header for debugging and API integration
# Core Concepts
Source: https://docs.decisional.com/core-concepts
The building blocks of Decisional
## Overview
Decisional is built around a few core concepts. Understanding these will help you get the most out of the platform.
The conversational operator for running agents and reviewing agent details
Durable AI workers that own repeatable workflows
The step-by-step plans your agents create and execute
External services where your team and agents get work done
Connections that expose app accounts, tools, and triggers
Every time your agent executes its workflow
Documents and files that give your agents context
Human review controls for sensitive actions and workflow gates
Where your team organizes agents and collaborates
***
## Dex
Dex is the conversational operator for Decisional. Use Dex to run agents, ask what happened in a run, view agent details, review approvals, or complete one-off tasks against the tools and context you have authorized.
Learn how Dex works across the web app, Slack, email, and mobile.
***
## Agents
Agents are the core operating units in Decisional. Each agent is an AI worker that owns a specific workflow and runs it for you.
You create an agent by describing what you want it to do in the Decisional web app. Once live, the agent runs from manual runs, schedules, triggers, or API calls, and you manage changes through the agent surfaces.
### What Makes Up an Agent
* **Instructions** — what you told the agent to do, written in plain English
* **Workflow** — the step-by-step plan the agent built to carry out your instructions
* **Integrations** — connections that give the agent tools and triggers for apps such as email, CRM, team chat, and spreadsheets
* **Sources** — documents and files that give your agent context
* **Triggers** — how your agent gets activated (schedule, event, or manual)
* **Approval policy** — when the agent pauses, asks for review, bypasses eligible approvals, or blocks an action
### Agent Operating Modes
| Mode | Purpose |
| ----------------- | --------------------------------------------------------------------------------------------- |
| **Chat Mode** | Talk with the agent, ask questions, run one-off tasks, and inspect context |
| **Operator Mode** | Configure instructions, workflow, tools, triggers, skills, sources, runs, and policy |
| **Run Mode** | Share a focused run page with people who should run or review work without changing the agent |
Understand Chat Mode, Operator Mode, and Run Mode for agents.
### Agent States
Your agent moves through these states as you set it up:
* **Draft** — still being configured or built
* **Ready** — tested and ready to go live
* **Live** — available for scheduled, triggered, API, or manual runs
* **Paused** — temporarily stopped, can be resumed anytime
***
## Workflows
When you describe what you want, your agent creates a workflow — a series of steps it follows to get the job done.
Each step in the workflow is a **node**. Nodes can do things like:
* Read data from a spreadsheet
* Call an API or integration
* Process a document
* Run AI reasoning
* Write results back
You do not need to draw every workflow step by hand. Decisional generates the graph from your instructions, then you review and change it in Operator Mode.
### Workflow Viewer
You can see your agent's workflow visually at any time. It shows every node, how they connect, and the status of each step during a run.
***
## Apps
Apps are external services where your team already works, such as Gmail, Slack, HubSpot, Google Sheets, or your own internal software. Agents can read from apps, take actions in them, use app events to start runs, and send completed work back to them.
In Decisional, you browse and manage apps through the **Integrations** catalog. Each catalog entry describes the app and shows its connection type, connected accounts, available tools, and triggers.
### Apps, Integrations, and Credentials
* An **app** is the external system where the work or data lives
* An **integration** defines how Decisional connects to the app and what agents can do
* A **connected account** or **credential** authorizes which account the integration can use
For example, Gmail is an app. The Gmail integration exposes tools for reading and sending email, triggers for sent or received messages, and connected accounts that determine which mailbox an agent can use.
***
## Integrations
An integration is Decisional's connection layer for an app. It brings the parts an agent needs into one place:
* **Connected accounts** — the authorized app accounts available in the workspace
* **Tools** — actions an agent can call, classified as read or write operations
* **Triggers** — app events that can start an agent, delivered by polling or webhook when supported
* **Approval policy** — workspace defaults and per-tool overrides that control whether a tool runs automatically or needs review
### How Integrations Work
1. Find the app in **Integrations** or let Dex identify it while building an agent
2. Connect an account using the app's supported authentication method
3. Review its available tools, triggers, and approval policy
4. Add the integration to an agent so its workflow can use those capabilities
The Integrations page also provides **Tool activity** for reviewing calls and a **Custom integrations** builder for apps that are not already in the catalog.
Open **Integrations** in the sidebar to browse apps, manage connected accounts, and review tool policies.
***
## Runs
Every time your agent executes its workflow, that's a run. You can see all your agent's runs from the **Runs** tab.
### Run Statuses
* **Running** — your agent is executing right now
* **Completed** — finished successfully
* **Failed** — something went wrong
* **Needs Approval** — your agent is waiting for you to approve an action
### What's in a Run
Each run shows you:
* A step-by-step breakdown of what happened at each node
* The input and output data for each step
* Files your agent produced
* How long it took
If a run fails, you can see exactly which step broke and why.
***
## Approvals
Approvals let agents pause before sensitive actions or manual decision points. A reviewer can see the context, approve the action, reject it, or route the workflow back for changes.
Approval behavior comes from:
* **Global approval mode** in Settings
* **Agent-level approval mode** for a specific agent
* **Tool approval policy** for connected tools and actions
* **Workflow gates** placed directly in a workflow
Use **Ask approvals** when you want a human review step. Use **Bypass approvals** only for stable workflows where eligible approval prompts can continue automatically.
Learn how approval modes, bypass approvals, tool policies, and global settings work.
***
## Sources
Sources are documents and files you upload to give your agents context. Think of them as reference material your agent can look up when it needs to.
### What You Can Upload
* **PDFs** — contracts, invoices, forms, statements
* **Spreadsheets** — price lists, templates, historical data
* **Word docs** — policies, procedures, playbooks
* **Images** — forms and receipts (your agent can read these with OCR)
### How Agents Use Sources
Decisional uses retrieval-augmented generation (RAG) to help your agent understand documents. When your agent needs information from a source, it retrieves only the relevant sections instead of trying to process the entire file. This works even across hundreds of pages.
***
## Workspaces
Workspaces are where your team organizes agents, integrations, and credentials.
### What's in a Workspace
* Your agents
* Connected integrations and credentials
* Team members and their permissions
* Sources and documents
### Team Access
Invite members from your organization into your workspace. Workspace owners control who can view, edit, and manage agents.
If you sign up with a company email, you'll automatically join your organization's workspace. Personal email addresses get their own workspace.
***
## How It All Fits Together
Describe what you want automated in the Decisional web app
Decisional creates an inspectable step-by-step plan and identifies the required tools
Upload any documents your agent needs to reference
Run a test, review each step, and set approval policy
Set a schedule, trigger, API call, public run link, or manual run path
Need changes? Open Operator Mode in the Decisional web app
***
## Next Steps
Build your first agent in minutes
Understand Chat Mode, Operator Mode, and Run Mode
Configure review, bypass, and global approval behavior
Deep dive into agent configuration and management
# Custom Integrations
Source: https://docs.decisional.com/custom-integrations
Turn API documentation and an API key into tools that Decisional agents can use
## Connect an API That Is Not in the Catalog
A custom integration lets Decisional agents work with an API that does not already have a built-in integration. Give Dex the API documentation, explain what the agent needs to do, and provide a test credential when the API requires one. Dex builds and tests the tools for you.
You do not need to write the integration code yourself. A clear API reference and a working way to authenticate are usually enough to get started.
Use a custom integration for an internal system, partner API, or less common service when you can provide one of these:
* A developer documentation URL
* An OpenAPI or Swagger specification
* A HAR export captured while using the service
* A cURL example or other request details
Custom integrations work best with documented HTTP APIs that Decisional can reach over the public internet. Private network addresses cannot be called from the builder.
## Build a Custom Integration
Open **Integrations**, choose **Custom Integrations**, and select **New session**.
Paste the developer documentation URL into the message box. If the documentation is in a file, use the **+** button to attach an OpenAPI specification or HAR export.
Tell Dex what the agent should read, create, update, or send. Name the few API actions that matter to your workflow instead of asking for every endpoint at once.
If the API requires authentication, provide a working test API key or the requested test account details. Dex uses it to replay a real request, confirm the authentication method, and test each tool.
Dex starts with a useful read-only action, tests its response, and then adds the remaining actions. Each action becomes a named tool with clear inputs and outputs.
When the integration card appears, select **Connect**. Enter the credential fields Dex configured, such as an API key, base URL, account ID, or region. This is the credential the agent will use after setup.
Select **Finish session** when the tools and credential setup are correct. The integration remains available on your organization's Integrations page and can be added to agents.
## What Dex Creates
Dex turns the API into the same building blocks used by other Decisional integrations:
| Part | What it does |
| -------------------------- | ------------------------------------------------------------------------------- |
| **Integration card** | Gives the API a name, description, and icon in the Integrations catalog |
| **Credential form** | Asks users for the values the API needs, such as an API key or account URL |
| **Tools** | Gives agents specific operations such as “Find a customer” or “Create an order” |
| **Validation check** | Checks whether a connected credential can reach the API |
| **Authentication refresh** | Can refresh supported short-lived authentication when the API requires it |
Read tools only look up information. Write tools create or change something in the external system and are marked separately so your normal approval policy can protect them.
Start small. Build and test one important read action before adding write actions. It is easier to confirm the data and permissions before the integration can change anything.
## Tell Dex Exactly What You Need
A useful request includes the documentation, authentication method, desired actions, and one safe example to test.
```text theme={null}
Build a custom integration for Acme CRM.
API docs: https://docs.example.com/api
Authentication: API key in the X-API-Key header
Create these tools:
1. Find a customer by email — read only
2. List open orders for a customer — read only
3. Add a note to a customer — write action
Use the test account and API key I provide. Test reads with customer
alex@example.com. Ask for approval before the write action.
```
If the documentation is incomplete, attach a HAR export recorded while you perform the action in the app. Dex can use the captured request to identify the endpoint, headers, and request body. Only capture traffic for the service you intend to connect.
## API Keys and Other Credentials
The builder needs a working test credential to prove that the integration works. Use a separate, low-permission test key when possible.
During the builder session, Dex uses the test value only to test requests and the authentication check. It does not place the secret in the tool code, the integration description, or the credential form definition.
After the integration is created, connect the account through the generated **Connect** form. Do not paste API keys into agent instructions or workflow nodes. The tools read the selected credential securely when they run.
Treat a HAR file as sensitive. It can contain cookies, authorization headers, and private request data. Use a test account, remove unrelated traffic, and revoke temporary credentials after setup when appropriate.
## Sessions and Sharing
Each builder session creates one integration.
| Session status | What it means |
| -------------- | ----------------------------------------------------------------------------------- |
| **Active** | You can select **Resume** and continue working with Dex |
| **Completed** | The session is read-only; select **View** to review the work |
| **Expired** | The builder workspace is no longer active; start a new session to continue building |
The session history is private to its creator. The custom integration itself belongs to the organization, so other organization members can find it on the Integrations page, connect their permitted credentials, and add its tools to agents.
## Use the Integration in an Agent
After connecting the integration:
1. Open an agent and unlock it to create an editable Draft if needed.
2. Ask Dex to add the custom integration and name the tools the workflow should use.
3. Open the integration in the right-hand panel and map a credential to each tool.
4. Test the agent with a small example.
5. Inspect the run and confirm each node received and returned the expected data before publishing.
Learn how integrations, tools, and credentials are shown and mapped in an agent.
Learn how connected accounts and tool credentials are managed.
# Dashboards and Usage
Source: https://docs.decisional.com/dashboards-and-usage
Monitor agent health, troubleshoot tool calls, and understand credit usage
## Operate Agents After They Go Live
Building an agent is only the beginning. Once agents are running on schedules, webhooks, or requests, use Decisional's dashboards to answer three different questions:
| View | Main question | Scope |
| ------------------- | ------------------------------------------------------------------ | ----------------- |
| **Agent Dashboard** | Are my workspace agents running successfully and on schedule? | Current workspace |
| **Tool Calls** | Which integration, toolkit, or tool is failing or taking too long? | Current account |
| **Usage** | Where are credits being spent, and should I set a cap? | Current account |
Start with the **Agent Dashboard** for overall health. Move to **Tool Calls** when an integration action is failing or slow. Use **Usage** to understand cost and control monthly agent spend.
The AI Usage and Tool Stats summaries on the Agent Dashboard are workspace-scoped. The full Usage and Tool Calls pages are account-wide. Check the account badge and use the same date range before comparing numbers across pages.
## Agent Dashboard
In **Agents** mode, select **Dashboard**. This is the day-to-day operating view for the selected workspace.
### What the Summary Cards Mean
| Card | What it tells you |
| ----------------------- | ------------------------------------------------------------------------------------------- |
| **Total Agents** | Number of agents in the workspace, with the number currently Live in parentheses |
| **Upcoming Runs** | Scheduled executions that are expected to start next |
| **Fixer-assisted runs** | Runs where Decisional's Fixer helped recover or repair execution during the selected period |
| **Completed Runs** | Runs that finished successfully during the selected period |
| **Failed Runs** | Runs that ended in failure during the selected period |
The **Agent Run Activity** chart shows completed, failed, and currently running executions over time. Use the 30-day, 14-day, 7-day, 24-hour, or custom date filter to separate a one-time problem from a repeated pattern.
Below the chart:
* **Published Agents** shows the Live and Paused agents in the workspace
* **Upcoming Runs** lists scheduled executions and their expected start time
* **Past Runs** links recent executions to their agent and run status
* **AI Usage** summarizes workspace model and Agent Node usage
* **Tool Stats** summarizes workspace tool volume and success
A small number of Fixer-assisted runs is not the same as a failed run—the Fixer may have recovered the execution. Repeated Fixer assistance on the same agent is still worth investigating because it can reveal a fragile instruction, changing data, or an unreliable integration.
### When Something Looks Wrong
1. Open the affected run from **Past Runs**.
2. Select the run on the workflow graph.
3. Find the first failed, incomplete, or unusually slow node.
4. If it is an integration action, open **Tool Calls** to compare it with other calls to the same tool.
5. If it is an AI or Agent Node step, open **Usage** to review model, Agent Node, and per-agent activity.
Learn how to inspect node execution, inputs, outputs, logs, and failures on the workflow run surface.
## Usage and Credits
Open the full **Usage** page from the **AI Usage** section on the Agent Dashboard. The account badge shows whether the numbers belong to an organization or a personal account.
### Balance and Spend
| Metric | Meaning |
| ----------------- | ----------------------------------------------------- |
| **Available** | Current credits available after active reservations |
| **Reserved** | Credits temporarily set aside for active metered work |
| **Total Balance** | Available and reserved credits combined |
| **Spent** | Credits actually charged during the selected period |
The **Daily Usage** chart groups credit-equivalent activity by resource type. Depending on the work performed, it can include:
* **AI / LLM Calls** for model requests
* **Agent Node** for sandboxed agent execution, including supported Claude Code, Codex, or Cursor runtimes
* **Platform Tools** for metered platform-provided integration calls
* **Workflow Runs**, **Agent Builds**, and **Storage** when those resource types apply
“Credit-equivalent” can include work paid for with a personal provider key. Personal-key usage is tracked so you can understand total activity, but it is not deducted from the Decisional credit balance. The **Spent** card and charged amounts exclude that personal-key portion.
### Understand What Is Driving Spend
Scroll down to compare models and agents.
Use **Spend by Model** to compare:
* Credits attributed to each model
* Number of model calls
* Input and output tokens
* Average credits per call
Use **Agent Spend & Caps** to compare:
* LLM spend and Agent Node spend for each agent
* Total calls and average credit use per model call
* Monthly cap usage, active reservations, and remaining credits
* Whether an agent is Live, at its cap, or blocked by its cap
You can search by agent name, UUID, or `agt_` public ID. The **All time** option changes the spend columns, but cap usage always represents the current UTC month.
### Set Monthly Agent Caps
**Agent Spend Default** sets the account-wide monthly cap for workflow-run credits. Individual agents can inherit this default or use their own override.
Enter a whole number of credits in **Monthly cap** and select **Save**. Leave the default clear when agents should be uncapped unless they have their own override.
In **Agent Spend & Caps**, compare each agent's current-month usage with its effective cap and remaining credits.
Check the agent's model calls, Agent Node activity, and recent runs. A sudden increase may come from more work, repeated retries, unexpectedly large inputs, or a workflow loop.
Update the account default when the policy should change for many agents. Use an individual override when one agent has a different workload.
The lower sections of the Usage page provide a full audit trail:
* **Usage by Resource Type** summarizes where credits went
* **Credit Grants** shows granted amounts, remaining balances, and expiration
* **Active Reservations** shows credits currently held for work in progress
* **Transaction Ledger** lists grants, debits, reservations, settlements, refunds, adjustments, and expirations
## Tool Calls
Open the full **Tool Calls** page from **Tool Stats** on the Agent Dashboard or from the Integrations area. Use it when a workflow reaches an integration step but the result is slow, missing, or incorrect.
### Read the Top-Level Metrics
| Metric | Meaning |
| -------------------- | ---------------------------------------------------------------------------------------- |
| **Total tool calls** | Number of recorded tool operations in the selected period |
| **Success rate** | Share of calls that completed successfully, with errors and timeouts shown underneath |
| **Avg latency** | Average call duration; p95 shows the duration that 95 percent of calls completed within |
| **Providers** | Number of tool providers with activity, with the busiest providers summarized underneath |
The **Tool Calls Over Time** chart separates successful calls from errors and timeouts. A short error spike can indicate a provider incident. A steady error rate on one toolkit is more likely to point to credentials, permissions, inputs, or an API change.
Continue down the page to compare:
* Calls, errors, and average latency by provider
* Top toolkits and individual tools
* Agents generating the most tool activity
* Recent individual tool calls
Use the filters to narrow recent calls by provider, toolkit, status, caller, agent, thread, or search text. This lets you move from an account-wide spike to the exact agent and tool involved.
Tool Calls covers recorded integration tools and supported Agent Node or orchestrator actions. Some internal workflow platform tools may be only partially represented, so the total is not necessarily the same as the number of workflow nodes executed.
## A Simple Weekly Review
Open the Agent Dashboard for each active workspace. Look for failed runs, recurring Fixer assistance, paused agents, and missed or unexpected schedules.
Open Tool Calls when errors increase. Narrow the data from provider to toolkit to tool, then identify the affected agent and recent call.
Open Usage and compare the current period with the previous one. Check the resource, model, and agent tables for unexpected changes.
Review the default monthly cap and individual agent overrides. Raise a cap only after confirming the higher usage is expected.
These views work together: the Agent Dashboard shows **what changed**, Tool Calls helps explain **why an integration step changed**, and Usage shows **what the work cost**.
# Working with Agents
Source: https://docs.decisional.com/guides/agents
Master agent creation, configuration, and optimization
## What Are Agents?
**Agents are AI workers** that execute business workflows. They follow your instructions, use configured tools and sources, and produce runs you can inspect.
Agents combine **your instructions** + **your tools** + **your sources** + **approval policy** to automate workflows while keeping runs reviewable.
## How an Agent is Organized
Agents are the main objects you create, operate, and share in Decisional. Dex can run agents and show details about their runs, while configuration changes happen in the Decisional web app.
Think about an agent in three layers:
The operating home for the agent. This is where you manage instructions, workflow, runs, outputs, inbox, state, activity, and context.
The material the agent can use while working. This includes files, sources, uploaded skills, and platform skills loaded for the job.
The ways people operate the agent: Chat Mode for conversation, Operator Mode for configuration, and Run Mode for public execution.
## Workspace Elements
The agent workspace is the control surface for a specific agent. The Agent Manager separates the agent's configuration, execution history, generated artifacts, inbox, state, activity, and context.
The plain-English operating spec for the agent. Instructions cover the job, success criteria, edge cases, triggers, tools, and runtime parameters.
The inspectable graph of nodes the agent executes. The workflow shows the steps, tool calls, gates, branching, and generated outputs.
The execution history for the agent. Runs show status, logs, node activity, approvals, files, outputs, and failure details.
Files, structured results, and other artifacts produced by the agent. Use outputs to inspect what a run created or download generated files.
The agent's email-style intake and response surface when an agent has an inbox. Use it to review inbound requests and outbound replies.
Durable state the agent keeps between runs, such as remembered values, counters, intermediate records, or configuration used by the workflow.
A recent timeline of actions, test runs, completed runs, and other agent events. Use it to jump back into the latest work.
Approval behavior, tool permissions, public access, and other operating controls that govern what the agent can do automatically.
## Agent Context
Context is the knowledge and capability the agent brings into a run. Keep it focused so the agent can retrieve the right information and load the right skill for the job.
| Context | What it is for |
| ------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- |
| **Files and sources** | PDFs, documents, spreadsheets, links, and uploaded files the agent can search and use as reference material |
| **Platform skills** | Built-in skills for connected systems, such as listing files, downloading files, sending messages, or working with business records |
| **User skills** | Reusable instructions or API-specific guidance you upload to teach the agent a specialized task |
| **Connected tools** | Authorized tool instances and credentials the agent can call during a workflow |
| **Run files and outputs** | Files uploaded for a specific run and files generated by previous workflow steps |
***
## Agent Modes
Use the mode picker to match the agent surface to the job:
| Mode | Best for |
| ----------------- | ----------------------------------------------------------------------------------------- |
| **Chat Mode** | Conversational requests, one-off work, explanations, and follow-up questions |
| **Operator Mode** | Changing instructions, workflows, tools, triggers, sources, skills, and approval behavior |
| **Run Mode** | Sharing a public run page with people who should not edit the agent |
Configuration changes belong in Operator Mode. Run Mode is for running or reviewing a public agent, not rebuilding it.
Learn how agents work across Chat Mode, Operator Mode, and Run Mode.
***
## Agent Lifecycle
The app shows the public state of each agent:
| State | What it means |
| ---------- | --------------------------------------------------------------------------------------------- |
| **Draft** | The agent is still being configured, connected, built, or tested. |
| **Ready** | The agent has been tested and can be made live. |
| **Live** | The agent can run from manual starts, schedules, triggers, API calls, or public run links. |
| **Paused** | The agent is temporarily stopped. No scheduled or triggered runs execute until it is resumed. |
***
## Creating Your First Agent
From the home page, type what you want to automate in the prompt box. You can also upload files for context, such as PDFs, price lists, or templates.
**Example:**
```
I want to process incoming invoices from my email,
extract key details, and update my tracking system
```
Your agent may ask follow-up questions to fill in gaps, such as which source to use, what format you want output in, or how to handle edge cases.
Decisional writes up a set of instructions — a plain-English summary of exactly what your agent will do. Review them and edit anything that's off in the Decisional web app.
Your agent identifies the required tools and integrations. Open the connections panel to connect any apps it needs.
Build the workflow, then test it to make sure everything works. Action nodes can pause for approval depending on the agent and tool policy.
Start with a simple workflow, test it, then gradually add complexity. Use Operator Mode in the Decisional web app for configuration-level edits.
## Writing Effective Instructions
Instructions are the **most important part of your agent**. Write them like you're training a smart new hire.
### Best Practices
**Bad Example:**
```
Process invoices
```
**Good Example:**
```
You are an Invoice Processing Agent.
For each invoice PDF uploaded:
1. Extract: Invoice number, date, vendor, line items, total
2. Verify the total matches the sum of line items
3. Look up the vendor in the approved vendor source
4. Write the extracted data to the connected invoice system
5. If total > $10,000, flag for manual review
```
Tell your agent what "done" looks like:
```
Success means:
- All required fields are filled
- Total is verified
- Vendor is matched
- Output is saved to the connected system
```
Anticipate problems and tell your agent what to do:
```
Edge cases:
- If vendor is not found, flag it for review
- If invoice is duplicate, skip it and log to Errors sheet
- If PDF is unreadable, flag for manual review
- If total doesn't match, recalculate and note discrepancy
```
If you've uploaded context documents, tell your agent when to use them:
```
Reference the "Invoice Processing SOP.pdf" for formatting rules.
Use the "Vendor Catalog.xlsx" to validate vendor names.
```
Show your agent what good output looks like:
```
Example output row:
| Invoice # | Date | Vendor | Total | Status |
| INV-001 | 2025-01-15 | Acme Corp | $5,432.12 | Approved |
```
## Adding Sources
Sources provide **context** to your agents so they can make informed decisions.
### Types of Sources
Contracts, invoices, forms, reports, SOPs
Price lists, catalogs, historical data
Policies, templates, procedures
Receipts, forms, diagrams (OCR enabled)
Company websites, documentation, APIs
Code, configs, raw text data
### How to Add Sources
In your agent page, click **Sources** in the left sidebar
Click **Add Source** → **Upload File**
Drag and drop multiple files at once (up to 50MB per file)
Click **Add Source** → **Add Link**
Paste a URL. Decisional will fetch and index the content.
Give each source a descriptive name so you can reference it in instructions:
**Example:** "Q4 2024 Price List" instead of "pricelist\_final\_v3.xlsx"
## Connecting Integrations
Integrations let your agent **take actions** in the systems where work happens.
### Using Integrations in Instructions
Once connected, reference integrations in your agent instructions:
```
When a quote is approved:
1. Send an email via Gmail to the customer
2. Update the deal stage in HubSpot to "Quote Sent"
3. Post a notification to the #sales Slack channel
4. Save the quote PDF to Google Drive
```
Make sure you've **authorized** the integration in Settings → Integrations before referencing it in instructions.
***
## Setting Up Triggers
Triggers control **when your agent runs**.
### Trigger Types
**Click the Run button**
Best for:
* Testing new agents
* One-off tasks
* Ad-hoc workflows
**How to use:**
Click **Run Agent** or start a manual run from Run Mode
**Time-based automation**
Best for:
* Daily reports
* Weekly summaries
* End-of-month processing
**Examples:**
* "Every day at 9 AM"
* "Every Monday at 8 AM"
* "First day of each month at 10 AM"
* "Every 6 hours"
**How to set:**
1. Click the calendar icon in Instructions tab
2. Choose frequency
3. Set time and timezone
**Triggered by external events**
Best for:
* Real-time automation
* Responding to user actions
* Integration-driven workflows
**Examples:**
* When a new record is created in a connected system
* When an email is received
* When a form is submitted
* When a webhook is called
* When a file is uploaded to Google Drive
**How to set:**
1. Connect the relevant integration
2. Go to Triggers tab
3. Select event type
4. Configure conditions (optional filters)
**Programmatic execution**
Best for:
* Custom applications
* External systems
* Developer workflows
**How to use:**
Make a POST request to the Decisional API:
```bash theme={null}
curl -X POST https://api.decisional.com/v1/agents/{agent_id}/run \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"input": "your input data"}'
```
Contact support if you need programmatic access for this workflow.
***
## Testing Your Agent
Before going live, always test your agent thoroughly.
### Testing Workflow
Create a small dataset (5-10 rows) with:
* Typical cases
* Edge cases
* Known error cases
Start a manual run and review the run details, node logs, outputs, and approvals
Check that:
* All expected columns are filled
* Data is accurate
* Formatting is correct
* Edge cases are handled
If results aren't perfect, refine your instructions and test again
### Common Issues & Fixes
**Problem:** Some columns are empty or incomplete
**Solutions:**
* Make it explicit in instructions: "You MUST fill all columns"
* Provide examples of complete output
* Add a validation rule: "If any required field is missing, flag for review"
**Problem:** Agent does something unexpected
**Solutions:**
* Break down complex steps into simpler ones
* Add more examples
* Use numbered steps instead of paragraphs
* Be more explicit about what NOT to do
**Problem:** Agent says it doesn't have the required data
**Solutions:**
* Check that sources are uploaded and indexed
* Reference sources by name in instructions
* Verify the information is actually in the documents
* Check that source names, field names, and instructions match
**Problem:** Agent takes too long to run
**Solutions:**
* Reduce the number of sources (only upload what's needed)
* Process in smaller batches
* Remove large, irrelevant documents
* Simplify instructions (fewer steps)
***
## Advanced Agent Patterns
### Multi-Step Workflows
For complex workflows, break them into multiple agents:
Collects and validates input data
Performs the core task
Sends results to stakeholders
Connect them with triggers: Agent 1's completion triggers Agent 2, and so on.
### Human-in-the-Loop
For high-stakes decisions, add manual review:
```
If the quote total exceeds $50,000:
1. Mark status as "Needs Review"
2. Send a Slack notification to @sales-manager
3. Wait for human approval before proceeding
```
### Error Handling
Build resilience into your agents:
```
Error handling rules:
- If an error occurs, log it to the "Errors" sheet
- Send a notification to the admin
- Continue processing remaining items (don't stop)
- Retry failed items once before flagging
```
### Chaining with APIs
Use agents to orchestrate API calls:
```
1. Fetch customer data from Salesforce API
2. Enrich with data from Clearbit API
3. Update the CRM with combined data
4. Send personalized email via Gmail
```
***
## Monitoring & Optimization
### View Run History
Click **Runs** in the left sidebar
See all past runs with:
* Status (Running, Completed, Failed, Needs Review)
* Duration
* Timestamp
* Inputs and outputs
Click any run to see:
* Full execution log
* AI reasoning step-by-step
* Source context and generated outputs
* Error messages (if failed)
### Performance Metrics
Track your agent's performance:
* **Success rate**: % of runs that complete successfully
* **Average duration**: How long runs take
* **Error patterns**: Common failure reasons
* **Manual review rate**: How often human intervention is needed
### Optimization Tips
* Upload only necessary sources
* Process in batches
* Use lighter file formats
* Cache reference data in a connected system
* Add more examples
* Upload better source documents
* Add validation rules
* Test with edge cases
* Increase schedule frequency
* Use event-based triggers
* Process in parallel (multiple agents)
* Optimize data structure
* Add error handling rules
* Validate inputs first
* Use try-catch patterns
* Log everything for debugging
***
## Best Practices
**Start simple, iterate:** Don't try to build a perfect agent on day one. Start with the core workflow and add features over time.
**Test with real data:** Use actual examples from your business, not synthetic test data.
**Document your agents:** Add comments in instructions explaining why certain rules exist.
**Version control:** Before making major changes, duplicate your agent so you can roll back if needed.
**Monitor continuously:** Check run history regularly, especially in the first few weeks.
**Get feedback:** Have actual users test your agent and report issues.
***
## Next Steps
Learn how files and sources give agents context
Ask questions about agents, runs, and approvals
Follow step-by-step guides for common workflows
# Using Dex
Source: https://docs.decisional.com/guides/ai-assistant
Start work, ask questions, run agents, and review approvals from the web app, Slack, email, or mobile
Dex is Decisional's conversational operator. Use it to ask questions about agents and runs, start one-off work, run agents, view agent details, and review approvals across the tools and context your workspace has authorized.
## How Dex Runs
Each organization's Dex runs on its own dedicated orchestrator computer (VM). This long-running computer receives requests, keeps channel connections available, and carries out work using the tools and context your organization has authorized.
Open **Settings → Organization** to see whether the computer is running. Organization owners and admins can open, stop, start, or delete it. They can also choose the model and runtime settings. Some settings changes briefly restart the computer while the new configuration is applied.
## What You Can Ask For
Ask about an agent, source, run, file, approval, or output without opening every underlying screen.
Use an agent's available context and connected tools for work that does not need a reusable workflow.
Summarize what happened, identify the step that broke, and surface the relevant logs, files, or outputs.
Start an agent, check its details, or review recent runs without opening every underlying screen.
## Channels
Dex can meet you where the work starts. You can connect several channels at the same time, such as Slack and email for the organization and SMS or iMessage for your own account. Requests from every connected channel are routed to the same organization Dex.
Start new work, inspect runs, review files, and open the Decisional web app when configuration needs to change.
Chat with Dex from team conversations, ask for updates, and keep work moving without leaving the thread where the request started.
Send requests, receive summaries, and handle asynchronous follow-ups through email when a workflow needs a written trail.
Text Dex from your phone through a personal connection. You can optionally route that number to a specific agent.
Connect a Telegram bot so people can talk to the organization Dex from Telegram.
Connect Dex to supported chats, spaces, or channels used by your team.
Configure these connections in **Settings → Organization → Channels**. Expand a channel to connect it or review its setup.
### How channel scope works
The scope label tells you who uses the connection:
* **Org-wide** connections are shared by everyone in the organization. Slack, email, Telegram, Google Chat, and Microsoft Teams use this scope.
* **Personal** connections belong only to the signed-in account. SMS or iMessage uses this scope, and each account can connect one number.
The Settings tabs use the same idea:
* **Account** contains settings that affect only you, such as your profile and notifications.
* **Workspace** contains settings for the workspace selected in the top-left menu, such as members, access, and workspace keys.
* **Organization** contains settings shared across the organization, including the dedicated Dex computer, model and runtime, channels, and audit log.
The selected workspace determines your current context and whether you have permission to manage the organization. It does not create a separate Dex computer for every workspace: the dedicated computer belongs to the organization.
## Approval requests
When you ask Dex to perform a write action from Slack, such as creating a Linear issue, Dex can post an approval request before the action runs. The request summarizes the action, shows the app or record involved, includes an approval link, and keeps the review in the same thread as the original ask.
In this example, Dex waits to create the Linear issue until someone selects **Approve creation**. The expiration timestamp limits how long the approval link can be used, and the thread keeps the request, approval, and resulting artifact together.
Learn how ask, bypass, and tool policies control agent actions.
## Relationship to Agents
Dex is where you ask for work and review what happened. Agents are saved automations with their own instructions, connected tools, schedules, files, run history, and approval settings.
Use Dex to run an agent or inspect its details and runs. For configuration changes, Dex points you to the Decisional web app, where you can edit the agent's setup.
## Related docs
Understand Chat Mode, Operator Mode, and Run Mode for agents.
Learn how ask, bypass, and tool policies control agent actions.
# Credentials
Source: https://docs.decisional.com/guides/credentials
Understand integrations, tools, credentials, and access control
## What Are Credentials?
Credentials are the authorized accounts, tokens, or keys Decisional uses when an agent calls an external system.
They sit underneath integrations, tools, and user tool bindings:
| Concept | Meaning |
| --------------- | --------------------------------------------------------------------------------------------------------------------------- |
| **Integration** | The external system or toolkit, such as email, team chat, CRM, file storage, or a database |
| **Tool** | A specific capability inside that integration, such as send email, list files, update a CRM record, or post a message |
| **Credential** | The authorized account, token, API key, or connection a tool uses to perform the action |
| **User Tool** | The binding between a tool and a credential, with a display name, optional defaults, and active or pending connection state |
An integration can expose many tools. A tool can use multiple credentials when your workspace has more than one authorized account for the same capability.
For example, a Gmail integration can include tools for sending email, searching messages, and reading attachments. The send-email tool might be able to use a founder's account, a support inbox, or a shared operations inbox, depending on which credentials are connected and allowed.
```mermaid theme={null}
flowchart LR
Gmail["Gmail Integration"]
SendEmail["Tool: Send Email"]
SearchMessages["Tool: Search Messages"]
ReadAttachments["Tool: Read Attachments"]
FounderSend["User Tool: Send as Founder"]
SupportSend["User Tool: Send as Support"]
OpsSearch["User Tool: Search Ops Inbox"]
FounderCred["Credential: founder@company.com"]
SupportCred["Credential: support@company.com"]
OpsCred["Credential: ops@company.com"]
Gmail --> SendEmail
Gmail --> SearchMessages
Gmail --> ReadAttachments
SendEmail --> FounderSend
SendEmail --> SupportSend
SearchMessages --> OpsSearch
FounderSend --> FounderCred
SupportSend --> SupportCred
OpsSearch --> OpsCred
```
In this example, the same Gmail integration exposes multiple tools. The same send-email tool is usable through more than one User Tool because each binding points at a different credential.
## Credential Types
Decisional supports a few credential patterns:
| Type | What it is used for |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------ |
| **Connected account** | OAuth or account-based authorization for an integration account |
| **Manual credential** | API keys, basic auth, database credentials, or custom integration secrets entered through a credential form |
| **Platform credential** | A Decisional-managed credential for supported tools, metered through credits instead of using your own connected account |
Credentials have connection status:
| Status | Meaning |
| ------------- | --------------------------------------------------------------------- |
| **Active** | The credential is usable and can be bound to tools |
| **Expired** | The credential needs to be refreshed or replaced before tools can run |
| **Initiated** | The connection flow started but has not completed yet |
Only active credentials are offered when choosing a credential for a tool.
## How Tools Use Credentials
When an agent runs a workflow step that calls an integration tool, Decisional resolves three things:
1. **The integration** the workflow is using
2. **The tool** needed for that step
3. **The credential** allowed to execute that tool
The selected credential determines which external account the action happens through. This is why credential labels and workspace sharing matter: teams often connect multiple accounts for the same integration, and the agent needs to use the right one.
### User Tool Bindings
Decisional stores the exact tool-to-credential pairing as a User Tool. A User Tool can be:
* Bound to a specific integration credential
* Bound to a platform credential when the integration supports it
* Pending when the tool exists but no credential has been connected yet
User Tools can also carry default parameters. Defaults are useful when a tool should usually run with the same account, folder, channel, database, recipient, or other option.
### Multiple Credentials for One Tool
A single tool can be used with different credentials. For example, the same send-message tool can be bound to a support credential, an operations credential, or a personal credential.
In the app, credentials are shown with labels and account identifiers so builders can choose the correct account. If no label is set, Decisional falls back to the account identifier when available.
## Credential Sharing
Credential visibility depends on the workspace.
| Workspace type | Credential behavior |
| ---------------------- | ----------------------------------------------------------------------------------------------------- |
| **Personal Workspace** | Credentials are private to the user. Integrations connected here are not usable by other people. |
| **Shared workspace** | Credentials can be shared with the workspace so multiple people and agents can use approved accounts. |
Credentials are still governed by access controls. Sharing a credential does not mean every action is allowed automatically.
Credential ownership still matters:
* The owner can rename, refresh, update, delete, and change sharing for their credential.
* Other workspace members can use shared credentials when policy allows, but they do not own the secret.
* Public or unauthenticated views only receive sanitized connection state. They do not expose internal credential IDs or secret values.
## Access Control Policy
Decisional applies policy at the integration and tool level.
Policy controls:
* Which integrations are available in a workspace
* Which tools are allowed for an agent or workflow
* Which credentials a tool can use
* Whether a tool can run automatically, must ask for approval, can bypass eligible approvals, or is blocked
This lets a team allow read-only tools broadly while keeping sensitive write tools, external sends, record updates, or destructive actions behind stricter controls.
Tools also carry an access type:
| Access type | Typical use |
| ----------- | -------------------------------------------------------------------------------------- |
| **Read** | Search, list, fetch, inspect, summarize, or retrieve data |
| **Write** | Send, create, update, delete, publish, comment, or otherwise change an external system |
Read tools are usually safer to make broadly available. Write tools should be reviewed more carefully because they can affect customers, records, files, or other systems.
Learn how tool policy combines with agent approval settings and workflow gates.
## Security Model
Agent code does not receive raw credential values. Decisional resolves the tool and credential server-side, executes the external call through the credential service, and keeps secrets out of agent-visible state.
The execution path is:
1. The workflow calls a User Tool by ID.
2. Decisional resolves the integration, tool, provider, defaults, and credential binding.
3. The credential service decrypts or resolves the credential only for that outbound call.
4. The provider adapter executes the action against the external system.
5. The agent receives the result, not the secret.
Sensitive credential fields are omitted from API responses and logs. Credential material is encrypted at rest and decrypted only in memory during execution.
Learn how Decisional protects credentials and workspace-scoped access.
# Monitoring Runs
Source: https://docs.decisional.com/guides/runs
Follow agent activity, inspect workflow steps, and recover incomplete work
A **run** is one execution of an agent. The Runs view shows what is active, what completed, and where an incomplete run needs attention.
## Open Agent Runs
Open an agent and choose **Runs** to see its execution history. If the agent has no runs yet, start a test or live run first.
The Agent Runs page has three parts:
* Summary cards for **Total Runs**, **Active Runs**, **Completed Runs**, and **Failed Runs**
* An activity chart for the selected time period
* A table of individual runs
### Filter the run list
Use the controls above the table to narrow the list:
| Control | Available choices |
| --------------- | --------------------------------------------------------------------------- |
| **Time period** | 30 days, 14 days, 7 days, 24 hours, or a custom date range |
| **Run type** | All, Test, Live, and Action. Build runs are also visible to administrators. |
| **Status** | Any run status, including active, waiting, and final states |
The run table shows the information that is available for every execution:
| Column | What it shows |
| ------------ | -------------------------------------------------------------------------------- |
| **Run ID** | The public identifier for the run |
| **Type** | Test, Live, Action, or Build |
| **Status** | The current run state |
| **Started** | Start date and time, relative time, and the person who started it when available |
| **Duration** | Elapsed execution time |
| **Action** | Cancel or Resume when that action is available |
A schedule, webhook, API call, or manual request can start a run, but the run list groups executions by **Test**, **Live**, **Action**, and **Build** rather than by trigger source.
## Select a Run
Select a row to open that run in the agent workspace. Decisional places the workflow in run context and opens the run summary on the right.
### Read the workflow graph
The graph shows which workflow path ran and the execution state of each node. During a run:
* A running node shows active progress.
* A successful node is marked as completed.
* A failed or rejected node is marked as failed.
* A node can show several executions when it runs once for each item or row.
* A reused result is identified when a successful test execution was reused instead of running the node again.
Select a node to review its executions. If the node ran more than once, choose the individual execution you want to inspect.
### Read the run summary
The run summary shows:
* Run ID, type, and start time
* Overall progress across the workflow
* Total, succeeded, running, and failed step counts
* A shortcut to pending approvals when a step needs review
* A credit warning when a step is paused for credits
* Run output files grouped by the node that created them
Open **Run Outputs** to view or download an individual output file. Use **Back to chat** to return to the conversation without clearing the selected run.
## Inspect a Node Execution
Select a workflow node, open **Executions**, and choose an execution to see what happened at that step.
The execution panel can include:
| Section | What it contains |
| ------------------ | ------------------------------------------------------------------------------- |
| **Execution Info** | Node name and type, execution ID, status, duration, retry count, and timestamps |
| **Error Details** | The error returned by a failed execution |
| **Input Data** | The structured values the node received |
| **Output Data** | The structured values the node produced |
| **Live Logs** | Streaming or saved log lines, with an option to copy them |
| **Live Browser** | A live browser preview when a running browser task provides one |
| **Log Files** | Log artifacts that can be opened individually |
| **Approval Info** | The reviewer and notes for an approval decision |
| **Dex Status** | Follow-up state for a Dex node, including what it is waiting for when available |
Start with **Error Details**, then compare **Input Data** and **Output Data**. Use **Live Logs** or **Log Files** when the error message alone does not explain the failure.
## Run Statuses
Every run has a status that tells you whether it is queued, active, waiting for something, or finished.
| Status | What it means | What you can do |
| ----------------------- | -------------------------------------------------------------------- | ------------------------------------------------------------------------------ |
| **Pending** | The run was accepted and is waiting to start. | Open it to monitor progress or cancel an eligible Live or Test run. |
| **Running** | One or more workflow nodes are executing. | Watch the graph, inspect active nodes, or cancel an eligible Live or Test run. |
| **Pending approval** | A node is paused until a person approves or rejects it. | Open the approval from the run summary or node, then approve or reject it. |
| **Needs credits** | A node paused before a metered AI or code-agent call. | Add credits or raise the agent limit, then retry the affected node. |
| **Self-healing** | Decisional is diagnosing a failed execution and attempting a repair. | Follow its progress or cancel an eligible run. |
| **Waiting for input** | The repair process asked a person for clarification. | Return to chat, answer the question, or cancel an eligible run. |
| **Completed** | All required work finished successfully. | Review the graph, node executions, outputs, and files. |
| **Partially completed** | Some work finished, but one or more parts remain incomplete. | Inspect the affected node and Resume an eligible Live or Test run. |
| **Failed** | The run stopped because required work could not complete. | Inspect the failed execution and Resume an eligible Live or Test run. |
| **Cancelled** | A person stopped the run before it finished. | Review completed steps and any external actions before starting again. |
**Pending**, **Running**, **Pending approval**, **Needs credits**, **Self-healing**, and **Waiting for input** can still change. **Completed**, **Partially completed**, **Failed**, and **Cancelled** are final states.
## Run and Node Actions
### Cancel a run
Cancel is available for eligible **Live** and **Test** runs while they are active or waiting. It is not offered for Build or Action runs from the run list.
Cancelling stops remaining work. It does not reverse work that already completed, such as a message that was sent or a record that was updated.
### Resume incomplete work
Resume is available for eligible **Live** and **Test** runs in **Failed** or **Partially completed** status. It continues incomplete work instead of starting the entire run again.
### Retry one node
For a failed, retrying, or credit-blocked node execution, the execution panel can offer:
* **Retry** to run that node execution again
* **Edit Code & Retry** to adjust the node code for this job and retry it
Code entered through **Edit Code & Retry** applies only to that job. It does not update the agent's main workflow or other jobs.
## Investigate an Incomplete Run
Select the Failed or Partially completed run from Agent Runs.
Use the workflow graph and run summary to locate failed, rejected, waiting, or credit-blocked work.
Select the node, open **Executions**, and choose the relevant execution.
Read the error, input, output, logs, and approval or credit state shown for that execution.
Fix the underlying data, credential, integration, source, instruction, or workflow configuration.
Retry the affected node when the execution panel offers it, or Resume the run when incomplete Live or Test work remains.
## Related Docs
Understand the graph and the nodes that execute during a run.
Learn how review steps pause and continue a run.
Monitor activity, tool calls, and credit usage across agents.
Ask about agents and runs from a conversation.
# Working with Sources
Source: https://docs.decisional.com/guides/sources
Master document processing, RAG, and context management
## What Are Sources?
**Sources provide context to your agents** - they're documents, files, and links that help your agent make informed decisions.
Without sources, agents only know what's in their instructions. With sources, agents can reference thousands of pages of documentation, policies, examples, and data.
***
## Why Sources Matter
Upload company policies, SOPs, pricing guides - your agent becomes an expert
Show your agent past quotes, emails, reports - it learns your style
Monitor competitor websites, documentation, news - stay up to date
Agents cite specific pages and sections - you can verify their reasoning
***
## Types of Sources
Decisional supports a wide variety of source types:
### Document Sources
**Best for:** Contracts, invoices, forms, reports, manuals
**Supported features:**
* Text extraction (native PDF text)
* OCR (scanned documents and images)
* Tables and structured data
* Multi-page documents
* Bounding box citations
**File size:** Up to 50MB per file
**Example use cases:**
* Upload past quotes so agent can match your formatting
* Upload pricing catalogs for product lookup
* Upload contracts for data extraction
* Upload SOPs for policy compliance
**Microsoft Word (.docx, .doc):**
* Policies and procedures
* Templates
* Reports
**Microsoft Excel (.xlsx, .xls):**
* Price lists
* Historical data
* Product catalogs
* Reference tables
**PowerPoint (.pptx):**
* Training materials
* Process diagrams
**File size:** Up to 50MB per file
**Formats:** PNG, JPG, JPEG, GIF, TIFF
**Best for:**
* Receipts (with OCR)
* Forms (with OCR)
* Diagrams and charts
* Handwritten notes (limited OCR support)
**Features:**
* Automatic OCR (text extraction)
* Visual understanding (charts, diagrams)
* Quality enhancement
**File size:** Up to 20MB per image
**Formats:** TXT, CSV, JSON, XML, Markdown
**Best for:**
* Code snippets
* Configuration files
* Structured data
* Raw text data
**File size:** Up to 10MB per file
### Web Sources
**How it works:** Decisional fetches the webpage content and indexes it.
**Best for:**
* Company websites
* Documentation pages
* Blog posts
* Product pages
* Competitor pricing
**Features:**
* Automatic refresh (daily, weekly, or custom)
* Change detection (notify when page changes)
* JavaScript rendering (for dynamic sites)
**Example:**
```
Add: https://example.com/pricing
Refresh: Daily
Agent instruction: "Monitor competitor pricing and update our prices to match"
```
**How it works:** Decisional calls your API and uses the response as context.
**Best for:**
* Live data from internal systems
* Third-party API data
* Real-time information
**Authentication supported:**
* API keys
* Bearer tokens
* OAuth 2.0
**Example:**
```
Add API: https://api.example.com/customers
Auth: Bearer token
Agent instruction: "Fetch customer data from API before generating quote"
```
***
## Adding Sources to Your Agent
In your agent page, click **Sources** in the left sidebar
Click **Add Source** and select:
* **Upload File** (PDF, Word, Excel, images, text)
* **Add Link** (webpage URL)
* **Connect Integration** (Google Drive, OneDrive, etc.)
**For files:** Drag and drop or browse to select
**For links:** Paste the URL and set refresh frequency
**For integrations:** Authorize and select files/folders
Give it a descriptive name (e.g., "Q4 2024 Price List" instead of "pricelist\_v3\_final.xlsx")
**Why:** You'll reference this name in agent instructions
Decisional indexes your source. This takes:
* Small PDFs: 10-30 seconds
* Large documents: 1-3 minutes
* Web pages: 30-60 seconds
You'll see a progress indicator.
Tell your agent when to use this source:
```
Reference the "Q4 2024 Price List" to look up product prices.
```
***
## How RAG Works
Decisional uses **Retrieval-Augmented Generation (RAG)** - a state-of-the-art technique for document understanding.
### The RAG Pipeline
When you upload a document:
1. **Text extraction**: Extract all text from the document
2. **Chunking**: Split into semantic chunks (paragraphs, sections)
3. **Embedding**: Convert text into vector representations
4. **Indexing**: Store in a semantic search index
When your agent runs:
1. **Query**: Agent formulates a search query based on its task
2. **Search**: Find the most relevant chunks from your sources
3. **Ranking**: Rank chunks by relevance
4. **Selection**: Select top K chunks (typically 5-10)
Agent uses retrieved context:
1. **Context**: Relevant chunks are provided to the agent
2. **Reasoning**: Agent processes task using context
3. **Citation**: Agent notes which sources it used
4. **Output**: Agent generates result with citations
### Bounding Box Citations
For PDFs, Decisional provides **exact location citations**:
Which page was referenced
Exact location on the page (highlighted region)
The specific text the agent read
**Why this matters:**
* **Verification**: You can check the agent didn't hallucinate
* **Auditing**: Understand exactly what the agent learned
* **Compliance**: Prove decision-making is based on real data
Click any citation in the run logs to see the exact PDF location highlighted!
***
## Optimizing Source Quality
Better sources = better agent performance.
### Document Preparation Tips
**Best:** Native PDF with selectable text
**Okay:** Scanned PDF (Decisional will OCR it, but slower and less accurate)
**How to check:** Try to select text in the PDF. If you can, it's text-based.
**If you only have scans:**
* Use high-resolution scans (300 DPI minimum)
* Ensure good contrast
* Avoid skewed or rotated pages
**Remove:**
* Cover pages with no useful info
* Blank pages
* Unrelated appendices
* Repetitive boilerplate
**Why:** Reduces noise, improves retrieval accuracy, speeds up processing
**How:** Edit the PDF before uploading or use page range selection (coming soon)
**Good structure:**
* Clear headers and sections
* Numbered lists
* Tables with headers
* Consistent fonts
**Avoid:**
* Text in images (unless necessary)
* Watermarks that obscure text
* Heavy redactions
* Multi-column layouts (if possible)
**Instead of:** One 500-page policy manual
**Do this:** Separate PDFs for each policy (Pricing, Returns, Shipping, etc.)
**Why:**
* Faster processing
* More accurate retrieval (agent knows which document to search)
* Easier to update individual policies
**Exception:** If the document is highly interconnected, keep it as one file.
**Bad:**
* `document.pdf`
* `final_FINAL_v3.pdf`
* `IMG_2024.png`
**Good:**
* `Acme_Corp_Pricing_Policy_2024_Q4.pdf`
* `Invoice_Template_Standard.pdf`
* `Product_Catalog_Hardware.pdf`
**Why:** Helps you and the agent identify the right source
***
## Source Organization Strategies
### By Workflow Type
**Sources to upload:**
1. **Price Lists** (Excel or PDF)
* Current pricing
* Discount tiers
* Product catalog
2. **Quote Templates** (PDF)
* Past successful quotes
* Shows formatting, tone, structure
3. **Pricing Policy** (PDF or Word)
* Rules for discounts
* Approval thresholds
* Special terms
**Agent instruction:**
```
Reference "2025 Price List" for product prices.
Use "Quote Template" as formatting guide.
Follow rules in "Pricing Policy" for discounts.
```
**Sources to upload:**
1. **Vendor List** (Excel)
* Vendor names
* Contact info
* Payment terms
2. **Chart of Accounts** (Excel or PDF)
* GL codes
* Category mappings
3. **Processing SOP** (PDF or Word)
* Step-by-step instructions
* Validation rules
* Edge case handling
**Agent instruction:**
```
Extract invoice data using "Processing SOP" guidelines.
Validate vendors against "Vendor List".
Assign GL codes from "Chart of Accounts".
```
**Sources to upload:**
1. **Example Contracts** (PDFs)
* Similar contract types
* Shows where key terms are located
2. **Field Definitions** (Word or PDF)
* What each field means
* How to interpret clauses
3. **Legal Terms Glossary** (PDF)
* Define complex legal terms
**Agent instruction:**
```
Extract key terms from contract.
Reference "Field Definitions" for what to extract.
Use "Legal Terms Glossary" to understand complex clauses.
Compare structure to "Example Contracts".
```
**Sources to upload:**
1. **Company Website Links**
* Add URLs of target companies
* Agent scrapes key info
2. **Industry Reports** (PDFs)
* Market research
* Competitor analysis
3. **Ideal Customer Profile** (PDF or Word)
* Defines what good leads look like
**Agent instruction:**
```
For each lead, visit their website and extract:
- Company size
- Industry
- Products/services
- Key contacts
Score leads against "Ideal Customer Profile".
Enrich with insights from "Industry Reports".
```
***
## Advanced RAG Techniques
### Multi-Document Reasoning
Agents can combine information from multiple sources:
**Example workflow:**
```
Agent instruction:
1. Look up product specs from "Product Catalog"
2. Find pricing from "Price List"
3. Check availability from "Inventory Report"
4. Verify compliance with "Export Regulations"
5. Generate quote combining all information
```
The agent will retrieve relevant sections from each document and synthesize them.
### Temporal Sources
For time-sensitive information, set refresh schedules:
Add a URL (e.g., competitor pricing page)
Choose frequency:
* Hourly (for real-time monitoring)
* Daily (for regular updates)
* Weekly (for less frequent changes)
* Manual (refresh when needed)
Get notified when the page changes
**Use case:**
```
Monitor https://competitor.com/pricing daily.
If their price drops below ours, send Slack notification.
```
### Versioned Sources
Keep multiple versions of the same source:
**Example:**
* `Price List 2024 Q1.xlsx`
* `Price List 2024 Q2.xlsx`
* `Price List 2024 Q3.xlsx`
* `Price List 2024 Q4.xlsx` (current)
**Agent instruction:**
```
Use "Price List 2024 Q4" for new quotes.
Reference historical price lists to show price changes over time.
```
***
## Integration-Based Sources
Connect live data sources through integrations:
**Setup:**
1. Connect Google Drive integration
2. In Sources, click **Add from Google Drive**
3. Select files or folders
4. Choose sync frequency
**Features:**
* Two-way sync (changes in Drive appear in Decisional)
* Folder monitoring (new files auto-added)
* Permissions respected (only files you have access to)
**Use case:**
```
Monitor "Sales Templates" folder in Google Drive.
When new template is added, agent learns the new format.
```
**Setup:**
1. Connect Microsoft 365 integration
2. In Sources, click **Add from OneDrive**
3. Select files or SharePoint sites
**Features:**
* Corporate SharePoint support
* Team site access
* Version history
**Use case:**
```
Sync all files from "Legal/Contracts" SharePoint folder.
Agent stays up to date with latest contract templates.
```
**Setup:**
1. Connect Gmail or Outlook integration
2. In Sources, create an email rule
3. Define which emails and attachments to monitor
**Features:**
* Filter by sender, subject, labels
* Extract specific file types
* Archive processed emails
**Use case:**
```
When email arrives from vendor@supplier.com with subject "Invoice":
1. Extract PDF attachment
2. Add as source to agent
3. Agent processes invoice
```
**Setup:**
1. Connect database integration (PostgreSQL, MySQL, Snowflake)
2. Write a query to fetch data
3. Schedule query execution
**Features:**
* SQL query support
* Parameterized queries
* Scheduled refresh
**Use case:**
```sql theme={null}
SELECT product_id, product_name, price, stock
FROM products
WHERE active = true
Refresh: Every hour
Agent uses results as product catalog.
```
***
## Managing Sources
### Source Library
View all sources across your workspace:
Click **Sources** in the main sidebar (not agent-specific)
See all uploaded documents, links, and connected files
* Search by name
* Filter by type (PDF, link, integration)
* Sort by upload date, size, usage
Attach existing sources to new agents (no need to re-upload)
### Source Actions
**For files:** Upload new version (overwrites old one)
**For links:** Click **Refresh Now** to re-fetch
**For integrations:** Sync automatically based on schedule
All agents using this source immediately see the update.
**Warning:** Deleting a source affects all agents using it.
**To delete:**
1. Click the source
2. Click **Delete**
3. Confirm
**What happens:** Agents can no longer reference this source. Existing citations remain in run history.
See which agents use a source:
1. Click the source
2. View **Used By** tab
3. See list of agents
**Why useful:** Before deleting, check if any agents depend on it.
Retrieve the original file you uploaded:
1. Click the source
2. Click **Download Original**
**Why useful:** Recover original if you've lost it.
***
## Troubleshooting Sources
**Common causes:**
1. **File too large** (limit: 50MB for PDFs, 20MB for images)
* Solution: Split into smaller files or compress
2. **Unsupported format**
* Solution: Convert to supported format (e.g., DOC → DOCX)
3. **Corrupted file**
* Solution: Re-export or re-save the file
4. **Network timeout**
* Solution: Try again with better internet connection
**Debugging steps:**
1. **Check source is attached to agent**
* Go to Sources tab in agent page
* Verify source is listed
2. **Reference source in instructions**
* Explicitly mention: "Use \[source name] to find..."
3. **Verify source is indexed**
* Check for "Indexed" status (not "Processing")
4. **Check source quality**
* Is text extractable? Try selecting text in PDF
* Is content relevant to agent's task?
5. **Review run logs**
* See if agent attempted to search source
* Check retrieval results
**Problem:** Agent cites irrelevant sections
**Solutions:**
1. **Improve source quality**
* Remove irrelevant pages
* Split into focused documents
2. **Be more specific in instructions**
* Instead of: "Look up pricing"
* Do: "Find product X in 'Price List' and return the 'Unit Price' column value"
3. **Add examples**
* Show agent what good results look like
4. **Check for conflicting sources**
* Multiple sources with similar content can confuse agent
* Remove duplicates or be explicit which to use
**Problem:** Agent can't read scanned PDFs or images
**Solutions:**
1. **Use higher resolution** (300 DPI minimum)
2. **Improve contrast** (black text on white background is best)
3. **Straighten pages** (no skew or rotation)
4. **Avoid handwriting** (OCR works best on printed text)
5. **Try native PDF** (if possible, use text-based PDF instead of scan)
If still failing, contact support - we can help with challenging OCR cases.
***
## Best Practices
**Quality over quantity:** 10 highly relevant documents > 100 generic documents
**Name descriptively:** "2025\_Q1\_Pricing" > "doc\_v3\_final.pdf"
**Keep sources updated:** Set refresh schedules or upload new versions regularly
**Reference explicitly:** Tell agents which sources to use and when
**Verify citations:** Check agent's cited sources in run logs to ensure accuracy
**Organize logically:** Group related sources, delete outdated ones
**Test with edge cases:** Upload challenging documents to see how well RAG handles them
***
## Next Steps
Deep dive into retrieval techniques and optimization
Learn how to write instructions that use sources effectively
Connect Google Drive, OneDrive, and other source providers
Advanced patterns for document workflows
# Workspaces
Source: https://docs.decisional.com/guides/workspaces
Collaborate with your team and manage access
## What Are Workspaces?
A **workspace** is your team's shared environment in Decisional. It contains your agents, sources, integrations, runs, settings, and team members.
A workspace is like a **Slack workspace** or **Google Workspace** - a shared space where your team collaborates on automation.
***
## Workspace Types
### Personal Workspace
**Created automatically** when you sign up with a personal email (Gmail, Yahoo, etc.).
Individuals, freelancers, small projects
* You're the only member (Owner)
* All agents are private
* Personal integrations and credentials
* Individual billing
Integrations connected in a Personal Workspace are only usable by you. Other people cannot join that workspace to reuse those credentials.
**Example:**
* Email: `john.smith@gmail.com`
* Workspace: "John's Workspace"
* Members: Just you
### Organizational Workspace
**Created automatically** when you sign up with a company email domain.
Companies, teams, departments
* Multiple team members
* Shared agents and resources
* Team integrations and credentials
* Centralized billing
* Audit logs
In any non-personal workspace, multiple people can collaborate on the same agents and share approved credentials. This lets a team connect the tools an agent needs once, then use those integrations according to the workspace roles and permissions.
**Example:**
* Email: `sarah@acmecorp.com`
* Workspace: "Acme Corp"
* Members: All employees with @acmecorp.com emails
If you're the first person from your company to sign up, you'll create the organization's workspace. Others who sign up with the same domain will automatically join.
Learn how integrations, tools, credentials, and access-control policy work across a workspace.
***
## Creating a Workspace
**Happens when you sign up:**
1. Visit [agents.decisional.com/sign-up](https://agents.decisional.com/sign-up)
2. Enter your email
3. Decisional detects your email domain:
* **Personal domain** (gmail.com) → Personal workspace
* **Company domain** (yourcompany.com) → Organizational workspace
4. Workspace is created automatically
5. You become the workspace **Owner**
**If your company already has a workspace:**
1. Sign up with your company email
2. Decisional detects existing workspace
3. You're added as a **Viewer** (default)
4. Workspace Owner can change your role
**Or get invited:**
1. Teammate sends you an invitation
2. Click the link in the email
3. Sign up or sign in
4. You're added with the role specified in the invitation
**Create multiple workspaces (Professional/Enterprise only):**
1. Click your profile → **New Workspace**
2. Enter workspace name
3. Choose workspace type
4. Click **Create**
**Use cases:**
* Separate workspaces for different teams
* Staging vs. production environments
* Client workspaces (agencies)
***
## Team Management
### Inviting Members
Click **Settings → Team** from the sidebar
Click the **Invite Member** button
Type the email address of the person you want to invite
Choose their role:
* **Owner**: Full control (only one allowed)
* **Admin**: Can manage team and agents
* **Editor**: Can create and edit agents
* **Viewer**: Read-only access
Click **Send Invite**
They'll receive an email with a link to join.
**Bulk invites:** Enter multiple emails separated by commas to invite many people at once.
### Managing Members
View and manage all workspace members:
| Column | Description |
| --------------- | --------------------------------------- |
| **Name** | Member's full name |
| **Email** | Their account email |
| **Role** | Current permission level |
| **Last Active** | When they last used Decisional |
| **Status** | Active, Pending Invitation, Deactivated |
| **Actions** | Change role, Remove, Resend invite |
**Actions you can take:**
* **Change role**: Click role dropdown → Select new role
* **Remove member**: Click ⋮ menu → Remove
* **Resend invitation**: For pending invites, click ⋮ → Resend
* **Deactivate**: Temporarily disable access without removing
Only **Owners** and **Admins** can invite and manage team members.
***
## Roles & Permissions
Decisional uses role-based access control (RBAC).
### Permission Matrix
| Permission | Owner | Admin | Editor | Viewer |
| ----------------------------- | ----- | ----- | ------ | ------ |
| **View agents** | ✅ | ✅ | ✅ | ✅ |
| **View sources** | ✅ | ✅ | ✅ | ✅ |
| **View run history** | ✅ | ✅ | ✅ | ✅ |
| **Run agents** | ✅ | ✅ | ✅ | ✅ |
| **Create agents** | ✅ | ✅ | ✅ | ❌ |
| **Edit agents** | ✅ | ✅ | ✅ | ❌ |
| **Delete agents** | ✅ | ✅ | ✅ | ❌ |
| **Upload sources** | ✅ | ✅ | ✅ | ❌ |
| **Manage integrations** | ✅ | ✅ | ✅ | ❌ |
| **Invite members** | ✅ | ✅ | ❌ | ❌ |
| **Remove members** | ✅ | ✅ | ❌ | ❌ |
| **Change roles** | ✅ | ✅ | ❌ | ❌ |
| **View billing** | ✅ | ❌ | ❌ | ❌ |
| **Manage billing** | ✅ | ❌ | ❌ | ❌ |
| **Change workspace settings** | ✅ | ❌ | ❌ | ❌ |
| **Delete workspace** | ✅ | ❌ | ❌ | ❌ |
| **Transfer ownership** | ✅ | ❌ | ❌ | ❌ |
| **View audit logs** | ✅ | ✅ | ❌ | ❌ |
### Role Descriptions
**Full control over the workspace**
**Responsibilities:**
* Manage billing and subscription
* Delete workspace (permanent!)
* Transfer ownership to another member
* Configure workspace-wide settings
* Approve major changes
**Limitations:**
* Only **one Owner** per workspace
* Cannot remove themselves (must transfer ownership first)
**Best for:** Workspace creator, executive sponsor, billing admin
**Can manage team and all resources**
**Responsibilities:**
* Invite and remove team members
* Create and configure agents
* Manage integrations
* View audit logs
* Handle day-to-day operations
**Limitations:**
* Cannot access billing
* Cannot delete workspace
* Cannot change workspace settings
**Best for:** Team leads, managers, power users
**Can create and modify agents**
**Responsibilities:**
* Build new agents
* Edit existing agents
* Upload sources
* Run agents
* Manage their own integrations
**Limitations:**
* Cannot invite team members
* Cannot view audit logs
* Cannot change workspace settings
**Best for:** Developers, automation builders, analysts
**Read-only access + can run agents**
**Permissions:**
* View all agents
* View sources
* Run agents
* View run history
**Limitations:**
* Cannot create or edit anything
* Cannot upload sources
* Cannot manage integrations
* Cannot invite team members
**Best for:** Stakeholders, executives, external collaborators
### Changing Roles
Settings → Team
Locate the person in the team list
Click the current role next to their name
Choose the new role from the dropdown
Changes take effect immediately
**Downgrading yourself:** If you're an Admin and downgrade yourself to Editor, you won't be able to change roles anymore!
***
## Workspace Settings
Configure workspace-wide options.
### General Settings
**Change your workspace display name:**
1. Settings → General
2. Edit "Workspace Name"
3. Click Save
**Example:** "Acme Corp" → "Acme Corp - Sales Team"
**Set timezone for scheduled agents:**
1. Settings → General
2. Select timezone (e.g., "America/New\_York")
3. Click Save
**Affects:**
* Scheduled run times
* Timestamps in logs
* Run history dates
**Upload custom workspace icon:**
1. Settings → General
2. Click "Upload Icon"
3. Select image (PNG, JPG, max 1MB)
4. Crop to square
5. Save
**Appears in:**
* Sidebar
* Workspace switcher
* Team member invitations
### Notification Settings
Configure workspace-wide notification preferences.
**Send email alerts to team members:**
**Options:**
* Agent run completed
* Agent run failed
* Agent needs review
* New member joined
* Billing updates
**Per-user override:** Members can customize their own notification preferences
**Post updates to Slack channels:**
**Setup:**
1. Connect Slack integration
2. Settings → Notifications → Slack
3. Select default channel (e.g., #automation-logs)
4. Choose events to post
**Events:**
* All agents completed
* Only failed agents
* Agents needing review
* Daily summary digest
**Call your API for workspace events:**
**Setup:**
1. Settings → Notifications → Webhooks
2. Add webhook URL
3. Select events
**Payload example:**
```json theme={null}
{
"event": "workspace.member_added",
"workspace_id": "ws_abc123",
"member": {
"email": "new.person@acme.com",
"role": "Editor"
},
"timestamp": "2025-01-15T14:30:00Z"
}
```
***
## Billing & Plans
Manage your workspace subscription.
### Viewing Billing
Settings → Billing (Owner only)
See your active plan:
* Plan name (Free, Professional, Enterprise)
* Monthly cost
* Included runs
* Billing cycle
Monitor current usage:
* Runs used this month
* Team members
* Active agents
* Storage used
### Upgrading/Downgrading
**Move to a higher plan:**
1. Settings → Billing → Upgrade
2. Choose new plan
3. Enter payment details (if not already on file)
4. Confirm
**Takes effect:** Immediately
**Billing:** Prorated charge for remainder of current period
**Move to a lower plan:**
1. Settings → Billing → Change Plan
2. Select lower tier
3. Review what you'll lose (e.g., extra team members)
4. Confirm
**Takes effect:** End of current billing period
**Refund:** No refund for remainder of current period
**Cancel subscription:**
1. Settings → Billing → Cancel Subscription
2. Select reason (helps us improve)
3. Confirm cancellation
**What happens:**
* Access continues until end of billing period
* All data remains accessible
* Can reactivate anytime
* After 90 days of inactivity, workspace is archived
**Before canceling:**
* Export run records
* Export agent configurations
* Save any important data
### Payment Methods
1. Settings → Billing → Payment Methods
2. Click "Add Payment Method"
3. Enter card details
4. Set as default (optional)
5. Save
**Accepted:**
* Visa, Mastercard, Amex, Discover
* Debit cards
* Corporate cards
**Change where invoices are sent:**
1. Settings → Billing → Billing Email
2. Enter new email
3. Verify email
4. Save
**Invoices sent to this email:**
* Monthly invoices
* Payment receipts
* Payment failures
**Access past invoices:**
1. Settings → Billing → Invoice History
2. See all past invoices
3. Click to download PDF
4. Filter by date range
**Includes:**
* Invoice number
* Date
* Amount paid
* Payment method
* Download PDF
***
## Security
### Workspace Security Settings
**Require 2FA for all workspace members:**
1. Settings → Security
2. Toggle "Require 2FA" ON
3. Set grace period (e.g., 7 days)
4. Save
**What happens:**
* All members must enable 2FA within grace period
* Cannot sign in without 2FA after deadline
* New members must enable 2FA before accessing workspace
**Available on:** Professional and Enterprise plans
**Auto-logout after inactivity:**
1. Settings → Security → Session Timeout
2. Set duration (15 min, 30 min, 1 hour, 4 hours, Never)
3. Save
**Applies to:** All workspace members
**Recommended:** 1 hour for high-security environments
**Restrict access to specific IP addresses:**
1. Settings → Security → IP Whitelisting
2. Add allowed IP ranges
3. Save
**Format:**
* Single IP: `203.0.113.0`
* Range: `203.0.113.0/24`
**Use case:** Corporate network, VPN-only access
**Available on:** Enterprise plan only
**Use your corporate identity provider:**
**Supported providers:**
* SAML 2.0
* OpenID Connect (OIDC)
* Okta
* Azure AD
* Google Workspace
**Setup:**
Contact [founders@decisional.com](mailto:founders@decisional.com) for SSO configuration
**Available on:** Enterprise plan only
### Audit Logs
Track all activity in your workspace (Professional/Enterprise only).
**What's logged:**
* Member added/removed
* Role changes
* Agent created/edited/deleted
* Integration connected/disconnected
* Source uploaded/modified
* Settings changed
**View logs:**
1. Settings → Audit Logs
2. Filter by:
* Date range
* Member
* Action type
* Resource (agent, source, integration, etc.)
3. Export to CSV
**Example log entry:**
```
[2025-01-15 14:30:22] sarah@acme.com
Action: agent.created
Resource: Quote Generation Agent (agent_abc123)
Details: Created new agent with Gmail integration
IP: 203.0.113.42
```
***
## Switching Workspaces
If you're a member of multiple workspaces:
Click the workspace name in the top-left corner
Choose which workspace to switch to
All agents, sources, integrations, and settings now reflect the selected workspace
**Each workspace is completely separate.** Agents, sources, and integrations are not shared between workspaces.
***
## Transferring Ownership
**Only the Owner can transfer ownership to another member:**
Settings → Team
Locate the member who will become the new Owner
Click ⋮ menu next to their name → Transfer Ownership
**Warning:** This cannot be undone!
Type "TRANSFER" to confirm
Ownership transfers immediately
You become an Admin automatically
**Cannot be reversed!** The new Owner must transfer it back to you if needed.
***
## Deleting a Workspace
**Only the Owner can delete a workspace. This is permanent!**
Settings → Danger Zone
Scroll to bottom → Delete Workspace
**What gets deleted:**
* All agents (permanently)
* All sources (permanently)
* All run history (permanently)
* All team members removed
* All integrations disconnected
**Cannot be recovered after 30 days!**
**Before deleting, download:**
* Agent configurations (Export as JSON)
* Run logs (if needed for compliance)
Type the exact workspace name to confirm
Workspace is immediately deactivated
Data retained for 30 days, then permanently deleted
***
## Best Practices
**Assign appropriate roles:** Give people the minimum permissions they need
**Enable 2FA:** Require it for all workspace members
**Regular audits:** Review team members quarterly, remove inactive users
**Separate workspaces:** Use different workspaces for staging and production
**Document processes:** Create internal docs for how your team uses Decisional
**Monitor usage:** Check billing dashboard monthly to avoid surprises
**Backup critical data:** Export agent configs, sources, and run records regularly
***
## Troubleshooting
**Possible causes:**
1. **Not an Admin/Owner:** Only Admins and Owners can invite
2. **Plan limit:** Free plan limited to 3 members
3. **Email already in use:** They're already in the workspace
**Fix:**
* Ask an Admin/Owner to invite
* Upgrade plan for more members
* Check team list for existing member
**Possible causes:**
1. **Wrong email:** Sign in with the email that was invited
2. **Invitation expired:** Ask for a new invitation
3. **Not yet accepted:** Check email for invitation link
**Fix:**
* Sign out and sign in with correct email
* Contact workspace Admin for new invite
**Possible causes:**
1. **Insufficient permissions:** Only Admins/Owners can change roles
2. **Trying to change Owner:** Owner role can only be transferred, not changed
3. **Changing your own role:** Can't promote yourself
**Fix:**
* Ask an Admin/Owner to change the role
* Use "Transfer Ownership" for Owner role
***
## Next Steps
Complete workspace setup walkthrough
Step-by-step invitation guide
Start building automation as a team
Configure 2FA and security features
# Integrations
Source: https://docs.decisional.com/integrations
Connect agents to the apps, triggers, and outputs used by Decisional templates
## Integrations power agent templates
Decisional integrations are not just app connections. They are the read tools, write tools, triggers, files, and approval surfaces that let an agent complete a workflow end to end.
The public template library shows the pattern clearly: each template combines a trigger, one or more connected systems, AI or data-processing steps, and a final output that is ready for review or downstream use.
At the platform level, each integration exposes tools, and each tool can use one or more credentials. Access-control policy decides which tools and credentials an agent can use and whether a tool call needs approval.
See the live template library on decisional.com.
Use Decisional-managed Apollo, Exa, Agent Mail, and Linq SMS tools without connecting your own provider account.
Turn API documentation and an API key into tested tools that your agents can use.
## What agents connect
Start agents from schedules, integration events, webhooks, form submissions, chat requests, or manual runs.
Pull context from inboxes, CRMs, spreadsheets, files, meeting notes, payment systems, issue trackers, and web sources.
Send messages, update records, create documents, write spreadsheets, archive files, comment on PRs, or notify reviewers.
Produce review packages, spreadsheets, documents, approval requests, summaries, exception notes, and audit-ready archives.
## Template-backed integration patterns
Use these public templates as examples of how integrations work together in real workflows.
| Template | Workflow area | Trigger | Connected systems | Output |
| ---------------------------------------------- | --------------------- | -------------------------------- | ------------------------------------------------------ | ------------------------------------------------ |
| **Stripe Invoice Processor Agent** | Finance | Scheduled | Stripe, Google Drive, Google Sheets | Invoice PDF archive and summary spreadsheet |
| **Stripe Outstanding Invoices Email** | Finance | On trigger | Stripe, Google Sheets, Gmail | Internal overdue invoice summary |
| **Procurement Approval Agent** | Finance & operations | Request intake | Google Sheets, Slackbot | Logged procurement request and approval flow |
| **Invoice Data Extraction Agent** | Finance | Email inbox | Gmail | Structured spreadsheet from invoice files |
| **Trade Finance Underwriting Agent** | Finance | Integration trigger | Gmail | Underwriting review workflow |
| **ETF News Summary Agent** | Wealth management | Scheduled or manual | Exa, Gmail, Slack, Google Drive | News summary and distribution package |
| **Attio Follow-up Automation Agent** | Sales | Scheduled | Attio, Google Sheets, Google Docs, Slackbot | Follow-up draft package with sheet and doc links |
| **HubSpot Follow-up with Specific Tone Agent** | Sales | Scheduled or task-based | HubSpot, Gmail, Google Sheets, Google Docs, Slackbot | Drafted follow-ups in the sender's tone |
| **HOT Lead Enrichment & Scoring Pipeline** | GTM | Integration trigger | HubSpot, Slack, Google Sheets, Gmail, Firecrawl | HOT lead alert and scored lead record |
| **Signed Up / Visited Email Sequence** | GTM | Product or visit event | Gmail | Four-step outreach sequence |
| **Email to HubSpot Sync Agent** | GTM | Email event | Gmail, HubSpot, Slackbot | Synced CRM contact and notification |
| **Market Data Enrichment Agent** | GTM research | Spreadsheet or manual run | Google Sheets, Exa, Firecrawl, Parallel, Apollo, Attio | Enriched lead and CRM record |
| **GitHub PR Review Agent** | Product & engineering | Slack request or new PR | GitHub, Slackbot | PR review comments and thread summary |
| **Weekly Shipping Report Agent** | Product & engineering | Scheduled | Linear, GitHub, Google Docs, Google Sheets, Slackbot | Weekly shipping report |
| **Call Recording Analyzer Agent** | Product & engineering | Scheduled or integration trigger | Fathom, Slackbot, GitHub, Linear | Product issue summary from calls |
| **Supabase Webhook Login Notifier Agent** | Product & engineering | Webhook | Slackbot | Signup notification |
| **Supabase User Onboarding Agent** | Product & engineering | Webhook | Slackbot, Attio | CRM update and onboarding notification |
| **Website Health Monitor Agent** | GTM | Scheduled | Ahrefs, Slackbot | Site health findings and alert |
| **HVAC Quoting Agent** | SMB operations | Form or manual intake | Google Sheets, Gmail | Professional quote |
| **Engineering Drawing Generator** | SMB operations | Email intake | Gmail, Google Drive | Generated drawing files |
| **Email Filtering Agent** | Email triage | Scheduled or manual | Gmail | Labeled and filtered inbox items |
## Common integration bundles
Templates usually combine several integration types rather than relying on one tool alone.
Finance templates often combine payment or invoice systems with spreadsheets, file storage, email, and approval messaging. This pattern is useful for invoice archiving, close support, procurement review, and exception handling.
GTM templates combine CRM records, inbox context, enrichment tools, spreadsheets, documents, and chat notifications. This pattern is useful for lead scoring, follow-up drafting, account research, and CRM hygiene.
Product and engineering templates connect repositories, issue trackers, call transcripts, chat, docs, and spreadsheets. This pattern is useful for PR review, shipping reports, call issue detection, onboarding, and webhook-driven notifications.
Operations templates combine email or forms with spreadsheets, file storage, and generated outputs. This pattern is useful for quotes, drawings, inbox routing, and lightweight back-office workflows.
## Trigger types
Templates use a small set of repeatable trigger patterns.
| Trigger type | Use it when | Example templates |
| --------------------- | ----------------------------------------------- | ---------------------------------------------------------------------------- |
| **Scheduled trigger** | The work should run on a cadence | Stripe Invoice Processor, Weekly Shipping Report, Attio Follow-up Automation |
| **Integration event** | A connected app event should start the workflow | HOT Lead Enrichment, Signed Up / Visited Email Sequence |
| **Webhook** | Another system should call Decisional directly | Supabase Login Notifier, Supabase User Onboarding |
| **Chat request** | A person should start work conversationally | GitHub PR Review Agent |
| **Manual run** | A user should run the agent on demand | HVAC Quoting Agent, Market Data Enrichment Agent |
| **Email intake** | New or recent email should be the input | Invoice Data Extraction, Email Filtering, Engineering Drawing Generator |
## How to choose integrations
Decide what the agent should produce: a spreadsheet, document, email, CRM update, approval request, archive, or notification.
Choose whether the work should start on a schedule, from an app event, from a webhook, from chat, from email, or manually.
Connect the systems the agent needs to read from, such as inboxes, CRMs, spreadsheets, issue trackers, meeting tools, payment systems, or files.
Connect the systems the agent needs to write to, such as chat, email, spreadsheets, documents, file storage, CRMs, or code review tools.
Use approval policy for write actions, customer-facing outputs, financial operations, and any step where a human should review before the agent proceeds.
Learn how tool policies, approval modes, and bypass approvals apply to integration actions.
Learn how integrations, tools, credentials, and access-control policy fit together.
## Connect integrations
You can connect tools ahead of time from the Integrations page in the app, or connect them during agent setup and agent runs when Decisional asks for access.
Start from a public template and connect the required systems during setup.
Describe the workflow in the Decisional web app, then connect the apps it asks for.
# Welcome to Decisional
Source: https://docs.decisional.com/introduction
Build and operate agents that work with the apps you use
## What is Decisional?
Decisional helps you build and operate AI agents for business workflows. You describe the work in natural language in the Decisional web app, and the agent builds, tests, runs, monitors, and repairs workflows across the apps and data you authorize.
**Describe what you want automated, then keep the work inspectable.** Agents build workflows, connect to tools, test before they go live, and expose runs, nodes, files, approvals, and tool activity so you can see what happened.
## Dex and agents
Dex is the conversational operator for your workspace. Use Dex for one-off work, questions, running agents, viewing agent details, and run review.
Agents are the durable workers. Each one owns instructions, workflow, tools, sources, triggers, runs, and approval policy.
## Automation Spectrum
Decisional addresses the full automation spectrum: Dex handles plain-English assistant work, Automation Agents run complex recurring processes, and each workflow can still use deterministic steps when a rule, trigger, approval, or API call is the right tool for the job.
Use Dex for one-off reasoning, drafting, questions, and work across the tools in your workspace. For recurring work, create a reusable agent in the Decisional web app.
Automation Agents build, test, run, monitor, and repair purpose-built workflows. They plan across steps, handle exceptions, recover from failures, and keep run history visible.
Not every step needs a model. Agents can use deterministic workflow steps for known paths, including triggers, schedules, API calls, file operations, approvals, and structured outputs.
## Why Decisional?
Describe what you want in plain English. Dex can handle one-off work, run agents, and show agent or run details.
Connect your business tools, files, and triggers. Agents use the systems you authorize.
Agents can build, test, run, monitor, and repair purpose-built workflows with visible run history.
Upload documents and spreadsheets as sources so agents can use the right context for the job.
## How It Works
Describe what you want automated in the Decisional web app. Upload relevant documents or spreadsheets for context.
Decisional creates an inspectable workflow, connects the right tools, and tests the workflow before it goes live.
See exactly what your agent will do before it goes live. Approve actions, set approval policy, or edit the workflow in the Decisional web app.
Your agent can run from triggers, schedules, public run links, or on demand. Monitor progress through runs, node logs, outputs, and chat.
## What Makes Decisional Different
Dex handles one-off work, agent runs, and run review. Agents can use connected tools, files, triggers, and skills to read, write, generate outputs, and keep work moving across business systems.
Decisional pairs AI judgment with inspectable workflow steps. Agents build and test workflows before they go live, show node-level execution, and can diagnose failed runs, retry transient errors, patch code-level issues, or ask for missing human input.
Every run leaves a trail. Teams can review run history, workflow steps, node logs, files, approvals, tool calls, usage, and agent-level credit activity, then adjust the automation from the same workspace.
## Who Uses Decisional?
Quote generation, invoice processing, reconciliation, and contract data extraction.
Outbound sequences, CRM updates, demo follow-ups, and lead enrichment.
Data entry, web scraping, email triage, and workflow orchestration.
Content pipelines, competitor analysis, and SEO workflows.
Portfolio monitoring, sign-up processing, and auto-responses at scale.
Custom workflows with the API and programmatic access.
## Example Use Cases
A few workflows teams run with Decisional:
* **Quote Generation**: Upload price lists, agent generates quotes automatically
* **Invoice Management**: Agent extracts contract data, creates invoices, sends via email
* **Data Entry**: Agent OCRs forms and receipts, populates your spreadsheet
* **CRM Updating**: Agent syncs meeting notes, updates deal stages automatically
* **Contract Extraction**: Agent parses PDF contracts, extracts key terms
## Get Started
Ready to build your first agent?
Create your first agent and run your first automation in minutes.
## Need Help?
Learn about Agents, Workflows, Integrations, and more
Learn when to use Chat Mode, Operator Mode, and Run Mode
Get help from our team
# Platform Integrations
Source: https://docs.decisional.com/platform-integrations
Use Decisional-managed research and communication tools without connecting your own account
## What Is a Platform Integration?
A platform integration uses an account managed by Decisional. It lets an agent use supported tools without asking you to connect your own API key or external account.
Look for the **Platform-provided** label when reviewing an agent's connections. The integration card shows whether calls use workspace credits and any current per-call charge.
Platform access may cover only part of an integration. For example, Decisional provides shared access to Apollo's read tools, but Apollo write actions require an Apollo account that you connect.
## Available Platform Integrations
| Integration | Use it for | Account setup |
| -------------- | --------------------------------------------------------------------- | ---------------------------------------------------------------------------- |
| **Apollo** | Finding and enriching people or companies | No account needed for supported read tools |
| **Exa** | Researching the web and finding related pages | No account needed for supported Exa search tools |
| **Agent Mail** | Giving an agent its own email inbox for outbound messages and replies | Create an inbox for the agent; no Gmail or Outlook connection needed |
| **Linq SMS** | Sending SMS or iMessage from an organization-assigned number | No OAuth connection, but Decisional must assign a number to the organization |
## How to Use a Platform Integration
Say what you want to find or send, what information Dex should start with, and what the final result should contain.
Dex adds the required tools to the agent. A **Platform-provided** connection is ready without an API key when every selected tool is covered by Decisional's account.
If a tool or trigger is outside the platform-provided scope, Decisional asks you to connect an account. This commonly happens when you want an Apollo write action or a trigger from an external app.
Run the agent with a small example. Select the run on the workflow graph, then click the research or communication node to review its inputs, outputs, logs, and credit usage.
## Apollo
Use Apollo when a workflow needs structured information about prospects, contacts, or companies.
### What You Can Do
* Search for people by details such as role, seniority, location, industry, or company
* Search for organizations and narrow the results to a target market
* Enrich a known person or company with additional Apollo data
* Enrich several people or organizations in one step
* Read supported Apollo records and reference data
The platform credential is **read-only**. Creating or updating Apollo records, changing sequences, or performing other Apollo write actions requires your own Apollo connection.
### How to Ask Dex
Give Dex clear search criteria and say which fields you want returned.
```text theme={null}
Use Apollo to find up to 25 VP or Head of Finance contacts at US manufacturing
companies with 100–1,000 employees. Return name, title, company, company domain,
location, LinkedIn URL, and work email when Apollo provides one. Do not guess
missing values.
```
For enrichment, provide a reliable starting point such as a company domain, LinkedIn URL, work email, or a person's name and company.
```text theme={null}
Enrich these contacts with Apollo. Match each person using their work email or
LinkedIn URL. Add current title, company, company domain, and location. Keep the
original row when Apollo does not find a confident match.
```
Apollo coverage varies. Treat missing data as unavailable and verify important contact details before using them in outreach. Platform-provided Apollo reads use workspace credits; the integration card shows the current charge.
## Exa
Use Exa when an agent needs current information from the public web rather than structured CRM data.
### What You Can Do
* Research a question and return an answer based on web sources
* Find pages related to a known article, company page, or URL
* Gather sources for market research, account research, news summaries, or competitive analysis
### How to Ask Dex
State the research question, the type of sources you want, and the final format.
```text theme={null}
Use Exa to research the most important announcements from Acme Corp during the
last 90 days. Focus on product launches, leadership changes, partnerships, and
funding. Return a five-bullet summary with the source URL for every claim.
```
To find related pages, give Dex the starting URL and explain what “similar” means for your task.
```text theme={null}
Use Exa to find pages similar to this customer case study. Prefer case studies
from B2B software companies that discuss finance automation. Return the title,
company, URL, and a one-sentence reason each result is relevant.
```
Exa is best for discovery and research. Use Apollo when you need structured people or company enrichment.
## Agent Mail
Agent Mail gives an agent a dedicated email inbox. Use it when the agent should send from its own Decisional-managed address instead of a person's Gmail or Outlook mailbox.
### How to Use It
1. Ask Dex to add Agent Mail to the agent.
2. Create or enable the inbox when the connection panel asks you to.
3. Tell Dex who should receive the message, the subject, and what the email should say.
4. If the workflow should continue after a reply, ask Dex to add an email reply trigger or an **Agentic Follow Up** Dex Node.
```text theme={null}
Use the agent's Agent Mail inbox to email each approved contact. Include the
account owner's name in the subject and attach the generated review file. Wait
for a reply and pass the reply to the next workflow step.
```
Use Gmail or Outlook instead when the agent needs a person's mailbox history, existing threads, or attachments stored in that mailbox.
## Linq SMS
Linq SMS lets an agent send SMS or iMessage from a phone number assigned to your organization.
### Before the Agent Can Send
1. Ask Decisional to assign a Linq number to your organization if one is not already available.
2. Add the recipient to the agent's SMS audience.
3. Ask the recipient to text the assigned number once. This changes the recipient from **whitelisted** to **subscribed**.
4. Send messages only after the recipient is subscribed.
```text theme={null}
Add these phone numbers to the coaching agent's SMS audience. List anyone who
has not subscribed yet. For subscribed recipients, send the approved coaching
message and wait for their reply before continuing.
```
Use phone numbers in international format, such as `+15551234567`. A whitelisted person cannot receive an outbound message until they have texted the assigned number and become subscribed.
## Platform Access and Your Own Accounts
You can connect your own account when you need capabilities that the platform credential does not cover. A user or workspace credential takes priority when selected for a tool.
Common reasons to connect your own account include:
* Using Apollo write actions
* Using a trigger tied to your external account
* Accessing private data that belongs to your organization
* Applying your own provider plan, limits, or permissions
Platform integrations can change as Decisional adds new managed services. The **Platform-provided** label in the Integrations catalog is the source of truth for what is currently available without account setup.
# Quickstart
Source: https://docs.decisional.com/quickstart
Build your first agent in minutes
## Before You Start
You'll need a Decisional account. [Sign up here](https://agents.decisional.com/sign-up) if you haven't already.
## Step 1: Describe What You Want
Go to [agents.decisional.com](https://agents.decisional.com) and sign in. You'll see a prompt box on the home page.
Tell Decisional what you want your agent to do in plain English. For example:
```
Every morning, check my CRM for new deals that closed yesterday,
pull the contract PDF from each deal, extract the key terms,
and add a row to my spreadsheet with the deal name, value, and renewal date.
```
You can also upload files for context — spreadsheets, PDFs, price lists, templates — whatever your agent needs to understand the job.
Be specific about what you want the outcome to look like. The more detail you give, the better your agent will be.
## Step 2: Review the Agent Plan
Once you submit, the setup flow asks clarifying questions to understand what you need, then identifies the integrations and tools required for the job.
After that, your agent writes up a set of **instructions** — a plain-English summary of exactly what it will do, step by step. You'll get to review these instructions before anything is built. If something's off, edit it in the Decisional web app.
## Step 3: Build Your Workflow
Once you approve the instructions, build the workflow. Decisional breaks your instructions into nodes, writes the logic for each step, and wires the graph together.
You can watch it build in real time from the **Workflow** tab. When it's done, you'll see the complete workflow laid out as a visual graph — each node representing a step your agent will execute.
Learn how trigger nodes, task nodes, action nodes, AI nodes, agent nodes, and gate nodes fit together.
## Step 4: Connect Your Integrations
Your agent needs access to the apps in your workflow. There are two ways to connect them:
* **Connections** — Open the connections panel on your agent page to see which integrations are needed and connect them directly.
* **Ask your agent** — Just tell your agent to connect an integration in the chat, such as "connect my email", and it will walk you through it.
Once you click connect, you'll authorize access through the app's standard OAuth flow. Decisional stores the connection and makes it available to the agent according to your workspace and tool policy.
Hover over any tool or trigger in the connections panel to see exactly where it's used in your agent's workflow.
You can also manage all your integrations from the **Integrations** page in the sidebar.
## Step 5: Test Your Agent
Before going live, run a test to make sure everything works.
Your agent will walk through its workflow step by step. For action nodes (things that write data, send emails, etc.), it pauses and asks for your approval before executing.
Learn when agents ask for approval, how bypass approvals work, and where global policy is configured.
Review the results, check the output, and if something needs tweaking — just tell your agent in the chat.
## Step 6: Go Live
When you're happy with the test, set your agent live. You can configure it to run:
* **On a schedule** — every morning, every hour, weekly, whatever you need
* **On triggers** — when a message, form submission, or connected-app event arrives
* **On demand** — run it manually anytime
## Step 7: Monitor and Manage
Once live, you can see all your agent's runs from the **Runs** tab. Each run shows:
* Status (Running, Completed, Failed, Needs Approval)
* What happened at each step
* The files and data your agent produced
Need to change something? Use Operator Mode in the Decisional web app for configuration-level edits.
## Next Steps
Learn about Agents, Workflows, Integrations, and Runs
Deep dive into agent configuration and management
Understand Chat Mode, Operator Mode, and Run Mode
## Need Help?
Get help from our team and other users on Discord
Reach out to our support team
# Frequently Asked Questions
Source: https://docs.decisional.com/reference/faq
Common questions about Decisional
## Getting Started
**No!** Decisional is entirely cloud-based. Just go to [agents.decisional.com](https://agents.decisional.com) in your web browser and sign up. No downloads, no installations.
**Recommended browsers:** Chrome, Firefox, Safari, or Edge (latest versions)
**About 5 minutes.** Our quickstart guide walks you through creating a Quote Generation Agent from start to finish. You'll:
1. Upload a spreadsheet (30 seconds)
2. Write instructions (2 minutes)
3. Test your agent (2 minutes)
Once you get the hang of it, you can create new agents in under 2 minutes.
**No coding required!** You write instructions in plain English, like training a new employee.
**Example:**
```
You are a Quote Generation Agent.
Look up prices from the price list and create quotes for customers.
```
However, if you want to use our API for advanced integrations, some programming knowledge helps.
**Yes!** Upload your existing Excel (.xlsx, .xls) or CSV files. All your data, formulas, and formatting are preserved.
You can also start with a blank spreadsheet and build from scratch.
**Your data is secure and private:**
* Stored in encrypted databases
* Hosted on secure cloud infrastructure
* Never shared with third parties
* You own your data and can export it anytime
Read our [Privacy Policy](https://www.decisional.com/privacy) for details.
***
## Agents & Workflows
Agents can handle a wide variety of business tasks:
**Document Processing:**
* Extract data from PDFs (invoices, contracts, forms)
* Generate documents (quotes, reports, emails)
* Classify and route documents
**Data Operations:**
* Update spreadsheets
* Enrich data from external sources
* Clean and validate data
* Perform calculations
**Integrations:**
* Send emails (Gmail, Outlook)
* Update CRMs (HubSpot, Salesforce)
* Post to Slack/Teams
* Query databases
* Call APIs
**Decision Making:**
* Route items based on rules
* Flag items for human review
* Prioritize and categorize
See [Workflows](/workflows/overview) for workflow structure and node types.
**Very accurate when properly configured.** Accuracy depends on:
1. **Clear instructions**: Specific, detailed instructions → better results
2. **Quality sources**: Clean, relevant documents → higher accuracy
3. **Good data structure**: Clean source data and clear fields → fewer errors
**Tips for maximum accuracy:**
* Test with real data before going live
* Validate input data before the agent runs
* Add verification steps in your workflow
* Review run logs regularly
For mission-critical tasks, use human-in-the-loop patterns (agent flags items for review).
**Yes!** Unlike traditional automation that breaks on edge cases, Decisional agents use **agentic reasoning** to adapt.
**Example:**
* **Traditional automation:** "If SKU not found, fail"
* **Decisional agent:** "If SKU not found, search by product name, suggest alternatives, or flag for review"
You guide this behavior in your instructions:
```
If the vendor is not in the vendor list:
1. Search for similar names
2. Add them as a new vendor
3. Send a notification to the admin
```
**Thousands of items in a single run.** Decisional is designed for batch processing.
**Typical performance:**
* Simple data updates: 1,000+ rows in 1-2 minutes
* Document extraction: 100+ PDFs in 5-10 minutes
* Complex workflows: Hundreds of items in under 10 minutes
**Best practices for large volumes:**
* Process in batches (e.g., 500 at a time)
* Use efficient source data structures
* Run during off-peak hours if time isn't critical
**Yes!** While an agent is running, you can:
* **Cancel**: Stop immediately (partial results may be saved)
* **Pause**: Coming soon
If you need to modify the agent, cancel the current run, make your changes, and start a new run.
***
## Sources & Documents
**Documents:**
* PDF (.pdf) - up to 50MB
* Microsoft Word (.docx, .doc)
* Microsoft Excel (.xlsx, .xls)
* PowerPoint (.pptx)
* Text files (.txt, .csv, .json, .xml, .md)
**Images:**
* PNG, JPG, JPEG, GIF, TIFF - up to 20MB
* OCR automatically applied
**Web:**
* Any public URL
* API endpoints (with authentication)
**Using RAG (Retrieval-Augmented Generation):**
1. **Upload**: You upload a PDF (e.g., pricing policy)
2. **Indexing**: Decisional extracts and indexes the text
3. **Retrieval**: When your agent runs, it searches for relevant sections
4. **Citation**: The agent references specific pages and paragraphs
5. **Verification**: You can click citations to see the exact source
**Bounding boxes** show you exactly what the agent read, highlighted in the PDF.
**Yes!** Decisional automatically applies OCR (Optical Character Recognition) to:
* Scanned PDFs
* Images of documents
* Receipts
* Forms
**For best results:**
* Use 300 DPI or higher scans
* Ensure good contrast (black text on white)
* Avoid skewed or rotated pages
* Use clean, printed text (handwriting is hit-or-miss)
**No hard limit**, but for best performance:
**Recommended:** 10-20 sources per agent
**Why?** More sources = longer retrieval time. Quality over quantity.
**Best practices:**
* Only upload relevant documents
* Remove duplicates
* Split large documents into focused files
* Archive outdated sources
**Yes!** Upload a new version with the same name to replace it.
**What happens:**
* Old version is archived (available in version history)
* New version is indexed
* All agents using this source automatically use the new version
**Use case:** Update your price list monthly - all agents instantly use new prices.
***
## Integrations & Triggers
**Email:** Gmail, Outlook, SMTP
**CRM:** HubSpot, Salesforce, Attio, Pipedrive
**Communication:** Slack, Microsoft Teams, Discord
**Storage:** Google Drive, OneDrive, Dropbox, Amazon S3
**Databases:** PostgreSQL, MySQL, Snowflake, BigQuery
**Payments:** Stripe, Bill.com
**Calendar:** Google Calendar, Outlook Calendar
**APIs:** Custom webhooks, REST APIs
**More coming soon!** Request integrations via [founders@decisional.com](mailto:founders@decisional.com)
**Yes.** You'll use OAuth to connect your existing accounts.
**Example:** To send emails via Gmail:
1. Click **Add Integration → Gmail**
2. Sign in with your Google account
3. Grant permission
4. Decisional can now send emails on your behalf
**Security:** You can revoke access anytime in your Google/Microsoft/etc. settings.
**Yes!** Once you connect Gmail or Outlook:
**Agents can:**
* Send emails to specific recipients
* Use templates
* Attach files (like generated quotes)
* CC/BCC
* Set custom subject lines
**Example instruction:**
```
After generating the quote, send an email via Gmail to the customer.
Subject: "Your Quote from Acme Corp"
Attach the quote PDF.
```
**Sent from:** Your email address (e.g., [you@company.com](mailto:you@company.com))
**Schedule your agent to run automatically:**
**Options:**
* Every X minutes/hours
* Daily at specific time (e.g., 9 AM)
* Weekly on specific days (e.g., Every Monday at 8 AM)
* Monthly on specific date (e.g., 1st of each month)
* Custom cron expressions (advanced)
**Time zones:** Specify your timezone to ensure correct scheduling.
**Example:** "Run every weekday at 9 AM Eastern Time"
**Yes! Use the Decisional API:**
```bash theme={null}
curl -X POST https://api.decisional.com/v1/agents/{agent_id}/run \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"input": "your input data"}'
```
**Use cases:**
* Trigger from your website when a form is submitted
* Integrate with your internal tools
* Build custom workflows
Contact support if you need API access enabled for your workspace.
***
## Pricing & Plans
**Yes! The free plan includes:**
* 100 agent runs per month
* 1 workspace
* Up to 3 agents
* Basic integrations
* Community support
**Perfect for:**
* Individuals
* Small projects
* Testing Decisional
**No credit card required** to sign up!
**Professional Plan:**
* 1,000+ runs per month
* Unlimited agents
* All integrations
* Priority support
* Advanced RAG features
* Audit logs
**Enterprise Plan:**
* Unlimited runs
* Dedicated infrastructure
* Custom integrations
* SSO (SAML, OIDC)
* SLA guarantees
* Dedicated support
Contact sales for pricing: [founders@decisional.com](mailto:founders@decisional.com)
**One run = one execution of your agent**, regardless of how many items it processes.
**Examples:**
* Agent processes 1 row → 1 run
* Agent processes 1,000 rows in same execution → 1 run
* Agent runs on schedule (even if no new data) → 1 run
**Tip:** Batch processing is efficient - process many items in one run!
**Yes!** Change plans anytime in Settings → Billing.
**Upgrading:** Takes effect immediately, prorated billing
**Downgrading:** Takes effect at end of current billing period
**Canceling:** You can export all your data before canceling.
***
## Troubleshooting
**Debugging steps:**
1. **Check run logs**
* Go to Runs tab
* Click the failed/unexpected run
* Read the AI reasoning to see what it thought
2. **Review instructions**
* Are they clear and specific?
* Did you provide examples?
* Are edge cases handled?
3. **Check sources**
* Are they attached to the agent?
* Are they indexed (not still processing)?
* Do they contain the expected information?
4. **Test with simple data**
* Try with 1-2 rows first
* See where it goes wrong
* Refine instructions
5. **Ask for help**
* Share run logs with our support team
* We can review and suggest improvements
**Common causes:**
1. **Too many sources** → Reduce to only essential documents
2. **Large sources** → Split into smaller, focused files
3. **Processing many rows** → Consider batching
4. **Complex instructions** → Simplify if possible
5. **Heavy integrations** → API calls add time
**Expected performance:**
* Simple tasks: Seconds per row
* Document extraction: 5-30 seconds per PDF
* Complex workflows: 30-60 seconds per item
If still slow, contact support - we can help optimize!
**Troubleshooting:**
1. **Check pop-ups** → Enable pop-ups for agents.decisional.com
2. **Check third-party cookies** → Enable in browser settings
3. **Check account permissions** → Make sure you're admin of the service
4. **Try different browser** → Sometimes browser extensions interfere
5. **Check OAuth app restrictions** → Your org may restrict OAuth apps
Still stuck? Email us the error message: [founders@decisional.com](mailto:founders@decisional.com)
**Possible reasons:**
1. **Removed by workspace owner** → Contact your workspace admin
2. **Email changed** → Sign in with your original email
3. **Workspace deleted** → Cannot be recovered (contact owner)
4. **Account locked** → Contact support
**Prevention:** Regular backups! Export agent configs, sources, and run records.
***
## Security & Privacy
**Yes! Security is our top priority:**
* **Encryption at rest**: All data encrypted in databases
* **Encryption in transit**: TLS/SSL for all connections
* **Access controls**: Role-based permissions
* **Audit logs**: Track who did what (Professional/Enterprise plans)
* **SOC 2 Type II**: Compliance in progress
* **GDPR compliant**: EU data protection standards
Read our [Security page](https://www.decisional.com/security) for details.
**Within Decisional:**
* **You and your workspace members** (based on roles)
* **Decisional engineers** (only for debugging, with your permission)
**Outside Decisional:**
* **No one.** We never share, sell, or access your data without permission
**AI models:**
* Your data is NOT used to train AI models
* Processed only for your specific workflows
**Yes**, with proper precautions:
**For sensitive data:**
1. Use **Enterprise plan** (dedicated infrastructure)
2. Enable **2FA** for all users
3. Use **audit logs** to track access
4. **Redact** sensitive info if not needed by agent
**Compliance:**
* **HIPAA**: Enterprise plan with BAA (contact sales)
* **GDPR**: Supported on all plans
* **SOC 2**: In progress
Email [founders@decisional.com](mailto:founders@decisional.com) for compliance questions.
**You have 30 days to recover:**
1. **Immediate:** Account marked for deletion, data inaccessible
2. **Day 1-30:** Contact support to restore
3. **Day 30:** Permanent deletion - all data erased
**Before deleting:**
* Download agent configs (JSON export)
* Back up sources
**Cannot be recovered after Day 30!**
***
## Still Have Questions?
Email us at [founders@decisional.com](mailto:founders@decisional.com)
Ask questions and share tips with other users on Discord
Get a personalized walkthrough
Explore in-depth guides and tutorials
# Glossary
Source: https://docs.decisional.com/reference/glossary
Key terms and concepts in Decisional
## Core Concepts
The conversational operator for Decisional. Dex helps you ask questions, run agents, inspect agent details and runs, review approvals, and complete one-off work.
**Example:** Ask Dex to run a quote review agent, then later ask it to explain a failed run or show the agent's approval settings.
An autonomous AI worker that executes your business workflows. Agents understand natural language instructions, process documents, and complete tasks automatically.
**Example:** A Quote Generation Agent that creates customized quotes based on customer requirements.
Documents, files, and links that provide context to your agents. Sources help agents make informed decisions by giving them access to company policies, examples, and reference materials.
**Example:** Uploading past quote PDFs so your agent can match your formatting style.
A single execution of your agent's workflow. Each time your agent performs its task, that's a run. Runs have a status (Running, Completed, Failed, Needs Review) and include detailed logs.
**Example:** When your Quote Generation Agent processes 10 customer requests, that's one run.
A container for your team's agents, sources, integrations, runs, and settings. Workspaces can be personal (individual use) or organizational (team collaboration).
**Example:** "Acme Corp Sales" workspace containing all quote generation and CRM automation agents.
A connection to an external service that your agent can interact with. Integrations enable agents to send emails, update CRMs, post to team chat, query databases, and more.
**Example:** An email integration allowing your agent to send quotes after approval.
The set of rules that decides whether an agent can proceed, must ask for approval, can bypass eligible approvals, or is blocked from taking an action.
**Example:** A reporting agent can always read records but needs approval before sending a final customer-facing message.
The mechanism that activates your agent. Triggers can be time-based (schedule), event-based (webhook, new data), manual (click Run button), or API-based (programmatic).
**Example:** "Run every day at 9 AM" or "Run when a new record is created in a connected system."
***
## Agent Modes
The conversational mode for an agent. Use it to ask questions, run one-off work, inspect context, and iterate with the agent.
The configuration and monitoring mode for an agent. Use it to change instructions, workflow, tools, triggers, skills, sources, runs, and approval behavior.
The focused run surface for an agent. When the agent is public, Run Mode gives people a public link to run or review the agent without accessing the full builder interface.
***
## Technical Terms
A technique where the AI retrieves relevant information from your documents before generating a response. This allows agents to reference specific pages, sections, and data points from your sources.
**How it works:**
1. Your documents are indexed when uploaded
2. When the agent runs, it searches for relevant sections
3. Retrieved content is used to inform the agent's decisions
4. The agent cites specific sources and page numbers
**Example:** Agent extracts pricing from "Q4 Price List.pdf" page 3, paragraph 2, and cites it in the quote.
The exact location within a PDF where information was found. Bounding boxes show the specific paragraph, table, or section the agent referenced, highlighted with a rectangular box.
**Why it matters:** You can verify the agent didn't hallucinate by seeing exactly what it read.
Technology that extracts text from images and scanned PDFs. Decisional automatically applies OCR to scanned documents so agents can read them.
**Best practices:**
* Use 300 DPI or higher for scans
* Ensure good contrast (black text on white background)
* Avoid skewed or rotated pages
The ability of AI agents to think through problems, handle edge cases, and adapt to new situations - not just follow rigid rules. Agentic reasoning allows Decisional agents to handle complex workflows that traditional automation can't.
**Example:** If a product is out of stock, the agent suggests alternatives instead of failing.
Search based on meaning, not just keywords. When your agent searches sources, it finds conceptually relevant information even if exact words don't match.
**Example:** Searching for "price" will also find "cost", "fee", "rate", etc.
***
## Workflow Terms
The data your agent receives to process. This typically comes from sources, API calls, email triggers, connected systems, or form submissions.
**Example:** Customer name, product, and quantity from a "Quote Requests" spreadsheet.
The results your agent produces. This can be sent through integrations, written to a connected system, saved as a file, or returned to another application.
**Example:** Generated quote with calculated prices, discounts, and total.
Natural language description of what your agent should do. Instructions are written like you're training a new team member.
**Example:**
```
You are an Invoice Processing Agent.
Extract key details from uploaded invoices and add them to the tracking spreadsheet.
```
A workflow pattern where the agent pauses for manual review before proceeding. Useful for high-stakes decisions.
**Example:** "If quote exceeds \$50,000, flag for manager approval before sending."
An approval mode where eligible approval prompts continue automatically instead of waiting for a human click. Blocked tools remain blocked.
**Example:** A stable internal report agent can bypass approvals for routine updates after it has been tested.
A workflow node that pauses execution for a human decision before the workflow continues.
**Example:** Require a manager to approve the final contract summary before the agent sends it.
Processing multiple items in a single run. Agents can process lists, records, files, or batches efficiently.
**Example:** Generate 100 quotes from 100 customer requests in one run.
How your agent responds to problems. Good agents log errors, continue processing remaining items, and notify you of issues.
**Example:** If one invoice fails to process, log it to an "Errors" sheet and continue with the rest.
***
## Integration & API Terms
A secure authorization method used by integrations. When you connect Gmail, HubSpot, etc., you're using OAuth to grant Decisional permission to access those services on your behalf.
A callback URL that triggers your agent when an external event occurs. Webhooks enable real-time automation.
**Example:** When a form is submitted on your website, it sends a webhook to trigger your agent.
A way for software to talk to other software. Decisional provides an API for programmatic access, and agents can call external APIs.
A secret token that authenticates API requests. Keep your API keys secure - they grant full access to your workspace.
A specific API URL that performs an action (e.g., `/agents/create`, `/runs/list`). Each endpoint has a specific purpose.
A standard way of building APIs using HTTP methods (GET, POST, PUT, DELETE). Decisional's API is RESTful.
***
## Status & States
The agent is currently executing. You can see real-time progress in the run logs.
The agent finished successfully. All outputs have been written and actions taken.
The agent encountered an error and couldn't complete. Check the run logs for details.
The agent is waiting for human approval before proceeding (human-in-the-loop pattern).
The item is waiting to be processed.
A source has been processed and is ready for the agent to search. Sources must be indexed before agents can use them.
A source is currently being indexed. Wait for this to complete before running your agent.
***
## Roles & Permissions
The workspace creator. Has full control including billing, deleting workspace, and managing all resources. Each workspace has exactly one owner.
Can manage all agents, invite team members, and configure integrations. Cannot access billing or delete workspace.
Can create, edit, and run agents. Cannot invite team members or delete workspace.
Read-only access. Can view and run agents, but cannot create or edit anything.
***
## File Formats
Simple text format where each line is a row and commas separate columns. No formatting, formulas, or multiple sheets.
**When to use:** Basic data export/import, no formatting needed
Modern Excel format supporting multiple sheets, formulas, charts, formatting, and more.
**When to use:** Full-featured spreadsheets with formulas and formatting
Older Excel format (pre-2007). Still supported but XLSX is preferred.
Static document format. Decisional can extract text and data from PDFs using RAG and OCR.
**Best for:** Contracts, invoices, reports, policies
Microsoft Word format. Good for policies, procedures, and template documents.
***
## Quick Reference
| Term | Category | Definition |
| ------------------- | ---------- | --------------------------------------------- |
| **Dex** | Core | Conversational operator for agents |
| **Agent** | Core | Autonomous AI worker |
| **Chat Mode** | Agent Mode | Conversational surface for an agent |
| **Operator Mode** | Agent Mode | Builder and control surface for an agent |
| **Run Mode** | Agent Mode | Focused public run surface |
| **Approval Policy** | Core | Rules for review, bypass, and blocked actions |
| **Source** | Core | Context document/file |
| **Run** | Core | Single execution |
| **RAG** | Technical | Retrieval-Augmented Generation |
| **OCR** | Technical | Optical Character Recognition |
| **Trigger** | Workflow | What activates the agent |
| **Integration** | API | Connected external service |
| **Owner** | Role | Full workspace control |
| **Indexed** | Status | Source ready to use |
***
## See Also
Learn about agents, sources, runs, and more
Common questions and answers
Detailed agent guide
Approval modes, bypass approvals, and global policy
Contact our team for help
# Security Policy
Source: https://docs.decisional.com/security
How Decisional protects agent execution, credentials, workspace data, production infrastructure, and customer data processing.
Decisional agents execute real code against real business data. The security architecture is built around enforced boundaries: sandboxed execution, credential isolation, workspace-scoped access control, and hardened production infrastructure.
Every control described here is implemented as an architecture-level constraint, not as a best-effort usage policy.
**No AI model training:** Decisional does not use client data, Customer Personal Data, prompts, files, integration records, agent inputs, or agent outputs to train AI models. Decisional also does not permit third-party model providers to train their models on customer data submitted through Decisional.
Every workflow execution runs inside short-lived compute boundaries, with additional isolation for code-execution steps.
Agent code calls tools through brokered IDs. Raw OAuth tokens and API keys never enter the sandbox.
Workspace data is scoped by `workspace_id`, backed by role checks and invite-gated onboarding.
Encryption, container hardening, deployment safety, rate limiting, and structured logging protect production systems.
## Sandboxed Execution
Decisional treats containment as a non-negotiable invariant. Each execution is enclosed by multiple boundaries so a failure in one layer does not propagate beyond it.
Each workflow execution receives its own short-lived runtime, created on demand and destroyed after completion. Runtimes are resource-capped, run without elevated privileges, have no persistent storage, and access platform data only through authenticated API calls.
Code-execution steps run in separate ephemeral compute environments with their own lifecycle. Sandboxes have hard timeouts, are destroyed on completion, and receive execution-scoped identifiers rather than credentials.
Workflow code runs with least-privilege access to authorized service interfaces only. It cannot directly access underlying credentials, platform databases, host infrastructure, or unrelated customer workspaces.
If an execution runtime crashes mid-run, the run resumes from the last checkpoint. Completed nodes are not re-executed, with node-level semantics backed by database checkpoints and durable artifact recovery.
## Trustless Credentials
The core credential guarantee is simple: agent code never sees raw credentials. Credentials are not placed in environment variables, function arguments, logs, or sandbox-visible state.
### Credential Storage
Connected-app access tokens and integration credentials are maintained by Composio, Decisional's integrations provider. Decisional stores brokered references to those credentials, not raw connected-app secrets, and agent code receives tool IDs instead of credentials.
For credentials and secrets Decisional stores directly, values are encrypted at rest with AES-256-GCM authenticated encryption. Key versioning supports rotation without downtime: old keys remain available for decryption, while new encryptions use the current key version.
Credential access is scoped to `workspace_id` and `user_id`, making credentials private by default with explicit sharing controls for team workflows.
### Brokered Proxy Pattern
Tool execution flows through a brokered chain where each layer sees only the minimum information it needs.
```text theme={null}
Agent code -> Routing service -> Composio / credential provider -> External API
tool ID credential refs decrypts in memory executes call
```
1. Agent code requests an authorized tool action using an internal tool identifier and structured parameters.
2. The routing service resolves the tool ID to provider configuration and credential references, never raw secrets.
3. Composio or the relevant credential provider is the decryption point for connected-app calls and holds decrypted material only as needed to execute the outbound call.
4. The provider adapter receives the live credential and executes against the upstream API. Raw connected-app credentials are not serialized into agent state, prompts, run logs, or sandbox-visible storage.
Are scoped to read-only operations. By default, read tools are allowed unless policy overrides require approval or block them.
Can require explicit human approval before actions such as sending email or updating records proceed.
Write approvals are recorded with acting identity, timestamp, run context, and operation metadata.
## Access Control & Multi-Tenancy
Each customer's data is isolated at the workspace layer. Agents, runs, credentials, files, schedules, and related entities carry a `workspace_id` foreign key, and protected data access is scoped through that workspace boundary.
### Role Hierarchy
| Role | View | Edit / Run | Manage Members | Delete Workspace |
| ------ | ---- | ---------- | -------------- | ---------------- |
| Viewer | Yes | No | No | No |
| Editor | Yes | Yes | No | No |
| Admin | Yes | Yes | Yes | No |
| Owner | Yes | Yes | Yes | Yes |
Every protected endpoint performs two checks: membership in the target workspace, then role-level permission for the requested operation.
New users cannot access protected workspace resources until invited by an Admin or Owner. Invitations use non-guessable, time-limited tokens.
Credentials are private by default. Workspace sharing is opt-in and enforced at every credential access point.
`dex_`-prefixed tokens are hashed before storage, scoped to specific permissions, and tracked by last-used timestamp.
Agents are private by default. Public run surfaces remain separated from protected workspace endpoints and are still governed by agent and tool policy.
## Infrastructure & Compliance
Production infrastructure enforces the lower-level controls that support the agent runtime, credential model, and tenant boundary.
Credentials are encrypted at rest with AES-256-GCM. Production database connections enforce TLS, and service-to-service calls use HTTPS.
Production containers use minimized images, non-root users, health checks, and path traversal protections.
Blue-green deployments validate health before traffic switches, with automatic rollback and graceful connection draining.
Distributed rate limiting protects public surfaces. Structured logs avoid credential values and record UUIDs plus metadata.
### Security Review Clarifications
| Review question | Decisional answer |
| -------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| What is the "Security Policy" referenced by the DPA? | This page is Decisional's Security Policy for the DPA. The public URL is [https://docs.decisional.com/security](https://docs.decisional.com/security). |
| Is there a SOC 2 report or trust center for integration credentials? | Connected-app access tokens and credentials are maintained by Composio, Decisional's integrations provider. Composio's Trust Center, including SOC 2 materials, is available at [https://trust.composio.dev](https://trust.composio.dev). |
| Are credentials encrypted? | Composio maintains integration credentials for connected apps. Decisional does not expose raw connected-app credentials to agents, sandboxes, prompts, logs, or run state. Credentials Decisional stores directly are encrypted at rest with AES-256-GCM, and service communication uses TLS/HTTPS. |
| Where is the current subprocessor list? | The maintained subprocessor list is the source of truth and is available at [Decisional Sub-Processors](https://app.notion.com/p/decisional/Decisional-Sub-Processors-12559fcfa75c80c7a954f514d08694b3). |
| Will customer data train AI models? | No. Client data, Customer Personal Data, prompts, files, integration records, agent inputs, agent outputs, and customer information are not used to train Decisional or third-party AI models. |
| Can customers object to material processing changes? | Yes. Decisional can add the material-change objection and termination language below to the DPA or customer agreement. |
| Can customers get specific deletion timing? | Yes. Decisional can add the 30-day during-term deletion timing language below to the DPA or customer agreement. |
### Data Processing Agreement
Decisional supports customer security and privacy reviews through a standard Data Processing Agreement based on the [Common Paper Data Processing Agreement Standard Terms v1.0](https://commonpaper.com/standards/data-processing-agreement/1.0).
The DPA has two parts: the customer-specific Cover Page and the Common Paper DPA Standard Terms. The Cover Page completes the agreement details, including the governing sales agreement, approved subprocessors, provider security contact, restricted transfer details, and the Annex I and Annex II processing information. If a signed DPA or customer agreement differs from this page, the executed agreement controls.
### DPA Clause Map
| DPA clause | What it means for Decisional customers |
| ----------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Cover Page, "Using this DPA" | The Cover Page controls if it conflicts with the standard terms. Capitalized terms are defined by the Cover Page, the DPA Standard Terms, or the customer agreement. |
| Cover Page, Annex II | Technical and organizational security measures are tied to the Security Policy referenced by the agreement. This page summarizes those controls. |
| Cover Page, Service Provider Relationship | Where CCPA applies, Decisional acts as a service provider for a limited and specified business purpose. Decisional does not sell or share Personal Data provided by the customer and does not retain, use, or disclose it except as needed to provide the service, as stated in the agreement, or as permitted by applicable data protection law. |
| Sections 1.1 and 1.2 | Decisional is a Processor when the customer is a Controller, and a Subprocessor when the customer is a Processor. |
| Sections 2.2 and 2.3 | Decisional processes Customer Personal Data only to provide and maintain the service, as configured through customer use of the service, as documented in the agreement, or under other written customer instructions acknowledged by Decisional. |
| Section 2.3 addendum | Customers may request a right to object to material service updates that increase privacy, security, or compliance risk. Pending resolution, Decisional will not implement the change for that Customer Data if the addendum language below is included in the agreement. |
| Section 2.6 | Subprocessors must be approved, listed with identity, country, and processing tasks, and bound by written obligations that limit access and use to the subcontracted service obligations. |
| Section 3 | Restricted transfers are supported through appropriate safeguards, including EEA SCCs, the UK Addendum, and Swiss transfer adjustments where applicable. |
| Section 4 | Security Incident notice is provided without undue delay where feasible and no later than 72 hours after Decisional becomes aware of the incident. |
| Sections 5.1 through 5.3 | Customers can request information reasonably necessary to verify DPA compliance, including security reports where available and reasonable written security due diligence requests. DPA compliance records are maintained for 3 years after the DPA ends. |
| Sections 6.1 and 6.2 | Decisional assists with third-party inquiries, data subject requests, DPIAs, and DTIAs where required by applicable data protection law and consistent with the DPA. |
| Sections 7.1 and 7.2 | Customers can delete Customer Personal Data through service functionality. After DPA expiration, Decisional returns or deletes Customer Personal Data at customer instruction unless further retention is required or authorized by law. Customers may request the 30-day during-term deletion timing language below. |
| Section 9 | If documents conflict, the order of precedence is: EEA SCCs or UK Addendum, then the DPA, then the customer agreement. |
| Section 10 | DPA obligations continue until the customer stops transferring Customer Personal Data to Decisional and Decisional stops processing it. |
### Available DPA Addendum Language
Decisional can include the following language in a customer DPA or agreement when requested.
**Material processing changes**
Where Decisional updates the Service and, by notice to Customer, changes the categories of data subjects, personal data, special category data, frequency of transfer, nature or purpose of processing, or duration of processing, Customer shall have the right to object to any such change that materially increases privacy, security, or compliance risk to Customer or to data subjects. Customer must raise such objection in writing within thirty (30) days of receiving Decisional's notice. The parties shall negotiate in good faith to resolve the objection within thirty (30) days. If the parties are unable to resolve the objection within that period, Customer may terminate the affected Service (or this Agreement in whole) without penalty upon written notice, and Decisional shall return or delete the affected Customer Data in accordance with the Deletion and Return of Customer Data section. Pending resolution, Decisional shall not implement the change with respect to Customer Data.
**During-term deletion timing**
During the Term, Decisional shall delete Customer Data no later than thirty (30) days after such data is no longer required for the purpose for which it was processed, or earlier upon Customer's written instruction, except to the extent retention is required by applicable law.
Where the customer is a Controller, Decisional acts as a Processor for Customer Personal Data. Where the customer is a Processor, Decisional acts as a Subprocessor.
The standard service scope covers Decisional's hosted SaaS product, including the web app, AI agents, APIs, integrations, workflow automation features, Dex, and associated services.
Standard processing covers customer end users or customers, including user activity and analysis data such as device information, IP address, and location information.
The standard DPA does not contemplate processing GDPR Article 9 special category data unless separately agreed in writing.
### Processing Activities
Decisional processes Customer Personal Data only to provide, maintain, secure, and improve the contracted service according to customer instructions and the applicable agreement. The standard cover page describes the following processing details.
| Activity | Examples |
| --------------- | ----------------------------------------------------------------------------------------------------------- |
| Receiving data | Collection, access, retrieval, recording, and data entry |
| Holding data | Storage, organization, and structuring |
| Using data | Analysis, consultation, testing, automated decision-making, and profiling as configured through the service |
| Protecting data | Access restrictions, encryption, monitoring, and security testing |
For standard configurations, data transfer frequency is batch sync every 24 hours or as otherwise agreed with the customer. Processing continues for as long as needed to provide the service, follow documented customer instructions, or comply with applicable law.
### International Transfers
The standard DPA includes transfer mechanics for restricted transfers under GDPR and UK GDPR. Where required, the EEA Standard Contractual Clauses and UK Addendum are incorporated through the DPA and completed by the customer-specific Cover Page.
The standard Cover Page identifies the Netherlands for EEA transfers and England and Wales for UK transfers. Customer-specific agreements may change those details.
### Customer Data Use Limits
Decisional's standard DPA and service provider terms limit how Customer Personal Data can be used.
Client data, Customer Personal Data, prompts, files, integration records, agent inputs, and agent outputs are not used to train Decisional or third-party AI models.
Where CCPA applies, Decisional does not sell or share Personal Data provided by the customer.
Decisional does not retain, use, or disclose Personal Data except as necessary to provide the service, as stated in the agreement, or as permitted by applicable data protection law.
Processing follows the customer's documented instructions, service configuration, the agreement, and any additional written instructions acknowledged by Decisional.
### Security Incidents, Deletion, and Review
Upon becoming aware of a Security Incident, Decisional notifies the customer without undue delay when feasible and no later than 72 hours after awareness, then provides timely information and takes reasonable containment and investigation steps.
Customers can delete Customer Personal Data through service functionality. After DPA expiration, Decisional returns or deletes covered data at the customer's instruction unless further retention is required or authorized by applicable law.
Decisional responds to reasonable written security, privacy, due diligence, and audit questionnaire requests according to the DPA and applicable confidentiality obligations.
Where available and appropriate, Decisional provides security report summaries or additional information under confidentiality so customers can verify security controls.
For compliance reviews, contact [founders@decisional.com](mailto:founders@decisional.com).
### Sub-Processors
Decisional uses vetted sub-processors for categories such as cloud infrastructure, authentication, model execution, sandboxed code execution, integration connectivity, observability, analytics, and payments.
Approved subprocessors are listed in the applicable DPA cover page or maintained subprocessor list so vendor names, locations, and processing purposes can stay accurate as the service evolves. The maintained Notion list is the source of truth: [Decisional Sub-Processors](https://app.notion.com/p/decisional/Decisional-Sub-Processors-12559fcfa75c80c7a954f514d08694b3).
Under Section 2.6 of the reviewed DPA, Decisional:
* provides advance written notice of intended subprocessor additions or replacements before beginning use of the new subprocessor;
* gives customers information needed to exercise their objection rights during the advance notice period;
* keeps written agreements requiring subprocessors to access and use Customer Personal Data only as needed for the subcontracted obligations and consistently with the agreement;
* imposes GDPR Article 28-style data protection obligations on subprocessors where GDPR applies;
* shares subprocessor agreements on request where required by the DPA, subject to redactions for confidential information, business secrets, and personal data; and
* remains responsible for subcontracted processing obligations, including subprocessor acts and omissions.
If a customer objects to a new subprocessor, the reviewed DPA provides cure paths such as canceling use of the subprocessor for that customer's data, offering an alternative service path, taking corrective steps, or disabling the affected service feature. If the objection remains unresolved and no commercially feasible alternative exists, the DPA allows termination of the affected subscriptions, order forms, or usage with refund of applicable prepaid fees for unused periods.
## FAQ
This page is Decisional's Security Policy for the DPA and is available at [https://docs.decisional.com/security](https://docs.decisional.com/security). It summarizes Decisional's technical and organizational measures, including sandboxed execution, credential isolation, access control, encryption, subprocessors, incident response, deletion, and DPA commitments.
Connected-app access tokens and credentials are maintained by Composio, Decisional's integrations provider. Composio's Trust Center, including SOC 2 materials, is available at [https://trust.composio.dev](https://trust.composio.dev). Decisional can provide additional security due diligence responses under the applicable agreement and confidentiality obligations.
Connected-app access tokens and integration credentials are maintained by Composio. Decisional stores brokered references to credentials, not raw connected-app secrets, and agent code receives tool IDs instead of credentials. Raw connected-app credentials are not serialized into agent state, prompts, run logs, or sandbox-visible storage.
Composio maintains connected-app credentials for integration calls. For credentials and secrets Decisional stores directly, values are encrypted at rest with AES-256-GCM authenticated encryption. Production database connections enforce TLS, and service-to-service calls use HTTPS.
No. Client data, Customer Personal Data, prompts, files, integration records, agent inputs, agent outputs, and customer information are not used to train Decisional or third-party AI models.
The maintained Notion list is the source of truth for current subprocessors, including provider names, processing activities, locations, and implementation dates. It is available at [Decisional Sub-Processors](https://app.notion.com/p/decisional/Decisional-Sub-Processors-12559fcfa75c80c7a954f514d08694b3).
No. The maintained subprocessor list identifies each provider's role and processing activity. Service providers used only for business operations may receive limited account information and usage data rather than Customer Personal Data.
Yes, subject to the applicable DPA. Under Section 2.6 of the reviewed DPA, Decisional provides advance written notice of intended subprocessor additions or replacements, gives customers information needed to exercise objection rights, and works through the cure paths described in the DPA if a customer objects.
Yes, Decisional can add the material processing change language shown above to the DPA or customer agreement. The clause gives customers 30 days to object to changes that materially increase privacy, security, or compliance risk, requires good-faith negotiation, and allows termination of the affected service if the objection is unresolved.
The reviewed DPA requires deletion through service functionality and return or deletion at DPA expiration at customer instruction unless further retention is required or authorized by law. Decisional can also add the 30-day during-term deletion timing language shown above to the DPA or customer agreement.
The standard DPA does not contemplate processing GDPR Article 9 special category data unless separately agreed in writing.
Under Section 4 of the reviewed DPA, Decisional notifies customers without undue delay where feasible and no later than 72 hours after becoming aware of a Security Incident, then provides timely information and takes reasonable containment and investigation steps.
The standard DPA includes transfer mechanics for restricted transfers under GDPR and UK GDPR. Where required, the EEA Standard Contractual Clauses and UK Addendum are incorporated through the DPA and completed by the customer-specific Cover Page.
Contact [founders@decisional.com](mailto:founders@decisional.com) for compliance reviews, security due diligence, subprocessor questions, DPA edits, or security report requests.
# Setup
Source: https://docs.decisional.com/setup
Create your account and set up your workspace
## Create Your Account
Go to [agents.decisional.com/sign-up](https://agents.decisional.com/sign-up) and create an account with your Google account or email.
If you sign up with a company email, you'll automatically join your organization's workspace if one exists. Otherwise, a personal workspace is created for you.
## Set Up Your Workspace
### Invite Your Team
Invite members from your organization directly into your workspace from the settings page. Workspace owners control who can view, edit, and manage agents.
### Connect Integrations
Your agents will ask you to connect apps as they need them, but you can also set them up ahead of time.
Go to **Integrations** in the sidebar, pick the app you want to connect, and follow the authorization flow. Common first integration categories:
* **Team chat** — for notifications and triggers
* **Email** — for sending and reading messages
* **CRM** — for customer and deal data
* **Spreadsheets** — for tabular data and lightweight operations
Integrations can be shared across your workspace or kept private to specific agents.
### Configure Approval Policy
The Settings page is where admins define the default approval mode used when a runner is set to **Default**.
Set whether runners should ask for approvals by default or bypass eligible approval prompts when they use Default.
In each toolkit's tools view, review the read/write defaults and any per-tool overrides: always allowed, needs approval, or blocked.
In the prompt-box approval picker, choose Default, Ask approvals, or Bypass approvals for the runner you are about to use.
Start new agents with Ask approvals in test runs. For live runs that should pause, set the runner picker to Ask approvals before starting.
Learn how workspace defaults, runner picker settings, tool policies, and workflow gates work together.
## Troubleshooting
Make sure pop-ups are allowed for agents.decisional.com and third-party cookies are enabled. If it still doesn't work, try email and password signup instead.
Check that you have admin access to the service you're connecting and that pop-ups are enabled. Try a different browser or clear your cache if it's still not working.
Ask a teammate to send you an invitation, or make sure you're signing up with the same email domain as your team.
## Next Steps
Build your first agent in minutes
Learn about Agents, Workflows, and Integrations
# Workflows
Source: https://docs.decisional.com/workflows/overview
Understand how your agent's workflow is structured
## What Is a Workflow?
A workflow is the step-by-step plan your agent follows to complete a task. When you describe what you want and hit **Build**, Decisional turns it into a sequence of steps that run one after another.
Each step is a node. Each node is connected to the next step by an edge, which shows what happens next.
### Purpose of the Workflow Graph
The workflow graph turns the agent's instructions into a plan you can see and inspect. It helps you understand what the agent will do before it runs and what happened after a run finishes.
Use the graph to:
* Follow the order of steps from the trigger to the final action
* See how information moves from one step to the next
* Tell which steps use fixed code and which steps use AI
* Find where the workflow needs approval or is waiting for a person
* Identify the step that failed and inspect what it received and returned
In the image below, the **Workflow panel is on the right**. It shows how a scheduled trigger starts the workflow, data moves through each connected node, and the final Action Node uploads and emails the completed report.
Not every step works in the same way. Some steps run **code** to do predictable work, such as fetching records, calculating values, or formatting a report. Other steps use **AI** to understand information, make a judgment, write content, or complete an open-ended task.
A workflow can combine both. Code steps handle work that should follow the same rules every time, while AI steps handle work that needs interpretation or flexibility.
In this example, the **Research Contact** step is backed by code. The code defines what information the step receives, the instructions and connected tools it uses, and the result it must return. You can open a node to inspect how that individual step works.
## Nodes
Nodes are the building blocks of a workflow. Each node owns one piece of work: starting the run, reading or transforming data, calling an app, asking for a decision, or generating output.
When an agent runs, Decisional executes the nodes in order and passes data from one step to the next through the workflow edges. This keeps the agent's process inspectable: you can see what each step did, what it received, what it produced, and where the run paused or failed.
### Node Types
Workflows are made up of different node types, each handling a specific kind of work:
The starting point of every workflow. Defines how the agent gets activated — manual run, schedule, webhook, or an event from a connected app.
The core workhorse. Reads data, processes information, makes decisions, and produces output. Most of your workflow logic lives here.
Takes an action in an external app or system of record. Action nodes are flagged during testing so you can approve before they execute.
Pauses the workflow for a human decision. Use gates before final sends, publishes, record updates, or other steps that require sign-off.
Runs a bounded AI model step to classify, summarize, extract, transform, or generate content from inputs passed through the workflow.
Gives an open-ended step to an AI agent that can use tools, make decisions, and work through several steps. Agent Nodes include the Dex Node.
## Viewing Your Workflow
After your agent builds, click the **Workflow** tab to see the visual graph. Each node is displayed with:
* Its **name** and a short description of what it does
* Its **type** (trigger, task, action, AI, agent, gate)
* **Connections** showing the flow of data between steps
You can zoom, pan, and click on any node to see its details and the code it runs.
### Inspect a Run on the Graph
Select a run to see what happened on the workflow graph. The graph changes from a plan of what should happen into a record of what happened during that run.
The run surface on the right shows the run ID, when it started, overall progress, and the number of steps that succeeded, are still running, or failed. Open **Run Outputs** to see files and other results created by the workflow.
The graph shows the status of each node for the selected run. For example, a green outline and check mark mean the node completed successfully. A node can also show that it is running, failed, waiting for approval, skipped, or still pending. When a node processes several rows, its progress shows how many row executions completed.
Click a node to open its **Executions** view. If the node ran once, Decisional opens that execution directly. If it ran more than once, select an execution from the list. The execution details show:
* Whether the node succeeded, failed, or is still running
* When it started, when it finished, and how long it took
* The information the node received and the result it returned
* Live logs, error details, approval information, and retries when available
This lets you move from the run summary to the exact node execution that produced a result or caused a problem.
## Agent Nodes
An Agent Node gives one part of a workflow to an AI agent. Unlike a fixed step, the agent can decide which actions are needed to reach the goal.
This is useful for:
* **Working with tools** — read files, use connected apps, compare results, and decide what to do next
* **Computer work** — work with files, images, documents, and other information in a managed environment
* **Multi-step tasks** — break a large task into smaller actions and work through them
* **Problem solving** — investigate why something failed and try the next best fix
The Agent Node receives information from earlier nodes and passes its result to the next step when it finishes.
### Dex Node: Agentic Follow Up
A Dex Node is designed for **Agentic Follow Up**. This means Dex can take an action, wait for a reply or a later time, and continue the same work without making you restart it.
You give Dex a goal in plain English and choose which connected tools it can use. Dex works through the goal, keeps the same context across turns, and reports back when the work is complete.
In this workflow, the Dex Node handles the coaching conversation between two fixed Action Nodes. It can use the connected tools shown on the node, message each recipient, and then pass the result to the final wrap-up step.
#### What a Dex Node Can Do
* **Use connected apps** — work with the exact email, Slack, CRM, SMS, or other tools assigned to the node
* **Take several turns** — keep working when the goal needs more than one action or decision
* **Wait until later** — pause until a time Dex chooses, then continue the same task with its earlier context
* **Wait for a reply** — send a supported email or Slack message, pause the workflow, and continue when someone replies
* **Return a result** — pass a summary and useful output to the next workflow step
The rest of the workflow waits while the Dex Node is working or waiting. It continues only after Dex completes the step. You can also turn on **Require approval** if someone should review the step before Dex starts.
#### What You See During a Run
| Status | What it means |
| ------------------------- | ------------------------------------------------------------------- |
| **Dex queued** | The step is ready and waiting for Dex to start. |
| **Dex running** | Dex is actively working on the goal. |
| **Dex scheduled** | Dex paused until a later time and will continue automatically. |
| **Dex waiting for reply** | Dex sent a message and is waiting for a response before continuing. |
| **Dex completed** | Dex finished the step and passed its result forward. |
| **Dex failed** | Dex could not finish the step. The run shows the reason. |
**Example:** Ask Dex to email a new customer about unfinished onboarding. If the customer does not reply, Dex can follow up two days later. When the reply arrives, Dex reads it, takes the next appropriate action, and completes the node only when the follow-up is finished.
Use a Dex Node when the exact number of actions, timing, or replies cannot be known in advance. For a single fixed app action, use an Action Node. For a short task such as classification or summarization, use an AI Node.
## AI Nodes
AI Nodes run a bounded model step inside the workflow. Use an AI Node when the job is clear, the required inputs are already available, and the step does not need an agent harness or tool use.
This is useful for:
* **Classification** — categorize emails, tickets, or documents into types
* **Summarization** — condense long content into key points
* **Extraction** — pull structured data from unstructured text
* **Generation** — write responses, draft emails, or create content
The AI Node takes input from previous nodes and returns model output that flows to the next step.
## Gate Nodes
Gate Nodes are explicit human approval checkpoints in a workflow. They do not run code or produce new output fields. Instead, they pause the workflow before downstream nodes run, then pass the same input data forward after the gate is approved.
Use a Gate Node when the workflow needs a named person or group to make a decision before continuing, such as manager approval before sending a customer email, finance approval before updating a record, or legal approval before publishing generated content.
Gate Nodes are different from approval flags on Action Nodes:
| Approval type | How it works |
| ------------------------ | --------------------------------------------------------------------------------------------------------------------------------- |
| **Action Node approval** | A simple approval pause attached to a specific action, often used before a write, send, publish, or external-system update. |
| **Gate Node** | A standalone workflow checkpoint with designated owners, approval instructions, voting rules, and optional public approval links. |
### Configure What Approvers Review
Ask Dex to configure the Gate Node with the information a person needs to make the decision. It can use values created by the trigger or any earlier step in the workflow.
You can organize the review page in a few simple ways:
| Review section | Best for |
| --------------- | ---------------------------------------------------------------------------------------------- |
| **Field cards** | Important details such as a job number, customer, location, total, or status |
| **Table** | A list of repeated items, such as invoice lines with a description, quantity, rate, and amount |
| **Checklist** | A short set of conditions the approver should confirm |
For each item, you can choose a clear label, the value to show, and how it should look. For example, a number can be shown as currency, a date can be made easier to read, and a missing value can use a fallback such as “Not provided.” Only the fields configured for review are saved in the approval request.
You can also configure the approval email separately. Add a short introduction and select a few summary fields so the approver understands the request before opening it. The **Review** button opens the full approval page.
The full page can show the Gate title and instructions, the workflow run, grouped review fields, tables, approver status, and an optional note. The approver can then **Approve** or **Reject** the request.
### Gate Owners
Each Gate Node has one or more owners. Owners are the only people who can approve or reject that gate. The gate can be configured in two modes:
| Owner mode | What happens |
| --------------------------- | --------------------------------------------------------------------------------------------------------- |
| **Any owner can approve** | The first owner approval resolves the gate and the workflow continues. A rejection fails the gate. |
| **All owners must approve** | Every listed owner must approve before the workflow continues. A rejection from any owner fails the gate. |
Gate settings can also include approval instructions, which are shown to approvers with the request, and shareable public links so external approvers can review without signing in.
During build runs, Gate Nodes are skipped for observability. During test and live runs, they pause execution until the approval rule is satisfied or the gate is rejected.
## How Workflows Get Built
You do not need to design workflows manually. Here's the process:
1. **You describe** what you want in plain English
2. **Decisional generates** the workflow — choosing node types, wiring them together, and writing the logic
3. **You review** the workflow in the visual viewer
4. **You can modify** the workflow in Operator Mode
If you need to make changes after building, use Operator Mode in the Decisional web app for configuration-level edits.
## Action Node Approvals
Action nodes are special because they affect external systems or records. During testing and configured review flows:
* Your agent **pauses** before executing each action node
* You see a **preview** of what the action will do
* You can **approve** or **reject** each action individually
This gives you a review point before sensitive actions touch real data or apps.
Approval behavior is controlled by the agent's approval mode, workspace tool policy, and any explicit gate nodes in the workflow. A live agent can ask for approvals, bypass eligible approvals, or block actions depending on that policy.
Learn how Ask approvals, Bypass approvals, global settings, and tool policies work.