Intelligent Slack agent that transforms how teams capture, track, and act on meeting outcomes — automatically.
Every day, millions of meetings happen across Slack channels. Decisions are made, action items are assigned, and commitments are given — only to be lost in the noise of message history.
- 70% of action items from meetings are never tracked
- Teams spend 5+ hours/week searching for past decisions
- No one remembers who promised to do what by when
Meeting Insights Agent solves this by automatically capturing, summarizing, and tracking everything that matters from your Slack conversations.
Generates a structured summary from any channel's message history, including:
- Meeting overview (3-4 line summary)
- Action items with assignees and deadlines
- Key decisions made
- Pending questions that need answers
Scans channel messages and extracts every action item, commitment, and task — formatted as a clear checklist with owners and deadlines.
Mention the bot with any question about your channel's history:
- "What was decided about the payment gateway?"
- "Who has pending tasks this week?"
- "Summarize last week's design review"
The bot searches through channel history and provides AI-powered answers with context.
The bot proactively monitors every message in real-time. When someone writes a commitment like:
- "I'll fix the auth bug by Friday"
- "Action item: database migration by Wednesday"
The bot instantly detects it and posts a tracked action item card — so nothing slips through the cracks.
┌──────────────────────────────────────────────────────┐
│ SLACK WORKSPACE (UI Layer) │
│ /meeting-summary │ /action-items │ @mention │
└──────────────────────┬───────────────────────────────┘
│ Socket Mode (WebSocket)
┌──────────────────────▼───────────────────────────────┐
│ BOLT FRAMEWORK (Event Processing) |
│ Slash Commands │ Event Listeners │ Web API │
└──────────────────────┬───────────────────────────────┘
│ REST API Calls
┌──────────────────────▼───────────────────────────────┐
│ AI PROCESSING (Groq LLM) │
│ LLaMA 3.3 70B │ Real-Time Search API │
└──────────────────────┬───────────────────────────────┘
│ Read / Write
┌──────────────────────▼───────────────────────────────┐
│ DATA LAYER │
│ Channel Messages │ User Profiles │ Action Items │
└──────────────────────────────────────────────────────┘
| Technology | Purpose |
|---|---|
| Slack Bolt Framework | App framework, event handling, Socket Mode |
| Slack Web API | conversations.history, users.info |
| Slack Real-Time Search API | Message search & context retrieval |
| Groq AI (LLaMA 3.3 70B) | Summarization, action extraction, Q&A |
| Node.js | Runtime environment |
| Socket Mode | Persistent WebSocket connection |
- Node.js 18+
- A Slack workspace
- Groq API key (Get one free)
# Clone the repository
git clone https://github.com/svk8190/meeting-insights-agent.git
cd meeting-insights-agent
# Install dependencies
npm install
# Configure environment variables
cp .env.example .env
# Edit .env with your tokensSLACK_BOT_TOKEN=xoxb-your-bot-token
SLACK_SIGNING_SECRET=your-signing-secret
SLACK_APP_TOKEN=xapp-your-app-token
OPENAI_API_KEY=gsk-your-groq-key
PORT=3000- Create a new app at api.slack.com/apps
- Enable Socket Mode and generate an App-Level Token
- Add these Bot Token Scopes:
chat:write,channels:history,channels:readusers:read,app_mentions:read
- Create slash commands:
/meeting-summary,/action-items - Subscribe to events:
app_mention,message.channels - Install the app to your workspace
node app.jsUser types /meeting-summary → Bot analyzes channel messages → Returns structured summary with action items, decisions, and pending questions.
User writes: "I'll complete the API integration by Monday" → Bot instantly posts:
📋 Action Item Detected! • Complete API integration • Assignee: @user • I'll track this for you!
Meeting Insights Agent addresses a universal workplace problem:
- Time Saved: Teams save 5+ hours/week previously spent searching for meeting outcomes
- Accountability: Every commitment is automatically tracked — nothing falls through the cracks
- Accessibility: Natural language interface means anyone can use it without training
- Real-time: Action items are detected the moment they're written
New Slack Agent — Building a new Slack agent that automates workflows and surfaces intelligent insights.
- Jira/Trello integration via MCP Server for auto-ticket creation
- Weekly automated digest emails with pending action items
- Multi-channel search and cross-team insights
- Custom reminder scheduling for detected action items
- Slack Marketplace listing for enterprise distribution
svk8190 — Built for the Slack Agent Builder Challenge
MIT