Decision Site

Search documentation

Search for pages in the documentation

Quick Start

Build your first workflow in 10 minutes

In this guide, you'll build a working workflow that automatically sends a Slack message with an AI-generated meeting summary when a meeting ends. This takes about 10 minutes.

What You'll Build

text
Meeting Ends ā†’ Load Meeting Data ā†’ AI Summarizes ā†’ Post to Slack

Prerequisites

Before starting, ensure you have:

  • Access to Decision Site with Creator or Admin role
  • Slack integration connected in your workspace (Setup Guide)
  • A Slack channel to post to

Step 1: Create a New Workflow

  1. From the dashboard, click New Workflow
  2. Enter a name: "Meeting Summary to Slack"
  3. Click Create

You'll see an empty workflow canvas with a node palette on the left.

Step 2: Add the Event Trigger

The trigger determines when your workflow runs. We'll use a meeting-ended trigger.

  1. In the node palette, find Triggers section
  2. Drag Event Trigger onto the canvas
  3. Click the node to open its configuration panel
  4. Set Event Type to MEETING_ENDED
  5. Click Save (or changes auto-save)

Your workflow now has an entry point that fires whenever a meeting ends.

Step 3: Add the Load Meeting Node

We need to fetch meeting details (title, attendees, transcript) before we can summarize them.

  1. In the node palette, find Loaders section
  2. Drag Load Meeting onto the canvas
  3. Position it to the right of the trigger
  4. Connect the trigger's success output to the Load Meeting's input:
    • Click and drag from the trigger's output port
    • Release on the Load Meeting's input port
  5. Configure the node:
    • Meeting ID: trigger.meetingPlanId (CEL expression)

The Load Meeting node now receives the meeting ID from the trigger and will fetch full meeting details.

Step 4: Add the AI Prompt Node

Now we'll use AI to generate a summary of the meeting.

  1. In the node palette, find AI section
  2. Drag AI Prompt onto the canvas
  3. Position it to the right of Load Meeting
  4. Connect Load Meeting's success output to AI Prompt's input
  5. Configure the node:

Model Tier: medium (good balance of quality and speed)

Messages:

Add a system message:

text
You are a helpful assistant that summarizes meeting transcripts concisely.

Add a user message (using Liquid template):

liquid
Please summarize this meeting:

Meeting: {{ json.meeting.title }}
Date: {{ json.meeting.startTime }}

Transcript Summary:
{{ json.callRecording.transcriptSummary }}

Provide a brief summary (2-3 sentences) and list any action items mentioned.

Return Type: string

Step 5: Add the Slack Post Node

Finally, we'll send the summary to Slack.

  1. In the node palette, find Actions section
  2. Drag Slack Post onto the canvas
  3. Position it to the right of AI Prompt
  4. Connect AI Prompt's success output to Slack Post's input
  5. Configure the node:

Channel: Enter your Slack channel ID or name (e.g., #sales-updates or C0123456789)

Message (Liquid template):

liquid
šŸ“‹ *Meeting Summary*

*{{ json.meeting.title }}*

{{ json.value }}

_Posted automatically by Agents_

Step 6: Review Your Workflow

Your completed workflow should look like this:

text
[Event Trigger] ā†’ [Load Meeting] ā†’ [AI Prompt] ā†’ [Slack Post]
   MEETING_ENDED    meeting data      summary       message

Checklist:

  • All nodes are connected
  • Trigger is set to MEETING_ENDED
  • Load Meeting references trigger.meetingPlanId
  • AI Prompt has system and user messages
  • Slack Post has channel and message configured

Step 7: Test Your Workflow

Before going live, test with the trigger execution feature:

  1. Click the Trigger button in the toolbar
  2. In the trigger execution modal, fill in the form with your test data (select a recent meeting)
  3. Click Execute

Watch the execution:

  • Each node shows status (waiting ā†’ running ā†’ completed)
  • Green checkmarks indicate success
  • Click any node to see its input/output data

Expected Result: A message appears in your Slack channel with the meeting summary.

Step 8: Release Your Workflow

Once testing succeeds, release your workflow:

  1. Click Release in the toolbar
  2. Confirm the release
  3. Your workflow is now active and will trigger automatically!

šŸŽ‰ Congratulations! Your workflow is now live. It will automatically run whenever a meeting ends.

Understanding the Data Flow

Let's trace how data moves through your workflow:

Trigger Output

json
{
  "meetingPlanId": "mp_abc123",
  "dealRoomId": "dr_xyz789",
  "event": "MEETING_ENDED",
  "firedAt": "2024-01-15T14:30:00Z"
}

After Load Meeting

json
{
  "meeting": {
    "id": "mp_abc123",
    "title": "Q1 Planning Review",
    "startTime": "2024-01-15T14:00:00Z",
    "attendees": [...]
  },
  "callRecording": {
    "transcriptSummary": "Team discussed Q1 goals...",
    "keyStatements": [...]
  }
}

After AI Prompt

json
{
  "type": "string",
  "value": "The team reviewed Q1 planning, focusing on revenue targets and product roadmap. Action items: 1) Sarah to finalize budget by Friday, 2) Mike to schedule follow-up with engineering."
}

Slack Post Result

json
{
  "success": true,
  "messageId": "1234567890.123456",
  "channel": "C0123456789"
}

What's Next?

Now that you've built your first workflow, explore these next steps:

Enhance This Workflow

  • Add error handling with an If node
  • Send to different channels based on meeting type
  • Include attendee names in the message

Learn More

Build More Workflows

Troubleshooting

Workflow doesn't trigger

  • Verify the workflow is Active (not Draft)
  • Check that meetings are being recorded in Decision Site
  • Review execution history for any failures

Slack message not appearing

  • Verify Slack integration is connected
  • Check the channel name/ID is correct
  • Ensure bot is invited to private channels

AI output is unexpected

  • Review the prompt for clarity
  • Check that meeting data is loading correctly
  • Try adjusting the model tier

General issues

  • Check the Troubleshooting Guide
  • View execution logs for detailed error messages
  • Contact support with execution IDs

You've completed the Quick Start! You now have a working automation and understand the basics of building workflows.

ā†’ Explore Node Reference to learn about all available nodes