How to Build an AI Agent in Slack Using Dumpling AI in make.com

Managing Slack messages manually can be overwhelming, especially when handling both text and audio-based instructions. With this automation in Make.com, we leverage Dumpling AI to create an AI agent that intelligently processes Slack messages. The workflow:

• Monitors new Slack messages in a specific channel.
• Routes messages based on type:
  • Audio messages: Transcribes the audio and performs a Google News search based on the transcribed text.
  • Text messages: Extracts key topics and performs a web scraping operation to gather relevant information.
• Posts the AI-generated results back into Slack.

This automation enables teams to stay informed with AI-driven insights directly within Slack.

Benefits of This Automation

• Automated Information Retrieval: Uses AI to pull real-time news and web data.
• Seamless Slack Integration: Automatically processes and responds to messages.
• Smart Routing: Dynamically handles audio vs. text messages differently.
• Saves Time: Eliminates the need for manual research and news tracking.

Step 1: Set Up the Slack “Watch Messages” Trigger

Purpose: Monitor a Slack channel for new messages.

1. In Make.com, create a new scenario and add the Slack app.
2. Select the Watch Messages module.
   • Connection: Link your Slack account (e.g., “My Slack (user) connection”).
   • Parameters:
     • Channel: team-building (or your target channel).
     • Limit: 2 (process two messages per trigger).
     • Input Method: Select from a list (choose channels easily).

Step 2: Route Messages Based on Content

Purpose: Split the workflow into two branches: one for files (e.g., audio) and one for text.

1. Add the Basic Router module.
   • This splits the flow into two routes using filters:
     • Route 1: Messages with files (e.g., audio/video attachments).
     • Route 2: Messages with text only.

File Route: Filter Condition: Ensure {{15.files[].mimetype}} exists.

Text Route: Filter Condition: Ensure {{Text}} exists.

Route 1: Process File Attachments

Step 2a: Download the File

1. Add the Slack: Download File module.
   • Connection: Use the same Slack account.
   • URL: Map {{15.files[].url_private_download}} (file URL from the trigger).

Step 2b: Transcribe Audio/Video with OpenAI Whisper

1. Add the OpenAI GPT-3: Create Transcription module.
   • Connection: Link your OpenAI account (e.g., “Make Partner Sandbox”).
   • File Data: Map {{13.data}} (downloaded file data).
   • Model: whisper-1 (for speech-to-text).
   • Response Format: Text (simple text output).

Step 2c: Generate an AI-Powered Response

Add the DumplingAI: Generate Agent Completion module.

Purpose: Fetch real-time news articles based on user queries (e.g., transcribed audio or text).

How It Works

1. Input Handling:
   • If a user uploads an audio file (e.g., a voice note requesting news), the workflow transcribes it using OpenAI Whisper.
   • If the input is text (e.g., “Latest updates on climate change”), the agent directly processes it.
2. Google News Search:
   • The AI agent parses the query (e.g., {{7.text}} from transcription or {{15.text}} from direct messages).
   • It searches Google News for relevant articles using keywords or phrases.
   • Example: A query like “AI breakthroughs in 2024” returns top news headlines, summaries, and links

• Connection: Link your DumplingAI account 
• Agent ID: a783c344-9bd6-4c30-8ec4-dffd987b5587 (replace with your agent’s ID) from Dumpling AI
• Messages: Map {{7.text}} (the transcribed text).
• Parse JSON: false (keep the response as plain text).

Step 2d: Post the Response to Slack

1. Add the Slack: Create Message module.
   • Connection: Use your Slack account.
   • Channel: team-building (same as the trigger channel).
   • Text: Map {{11.text}} (DumplingAI’s response).
   • Use Markdown: true (format the response clearly).

Route 2: Process Text Messages

Step 2a: Generate an AI Response Directly

Purpose: Extract specific data from websites based on user requests (e.g., product details, research data).

How It Works

1. Input Handling:
   • Processes text-based queries (e.g., “Scrape pricing from example.com” or “Extract stats from XYZ report”).
2. Web Scraping:
   • The AI agent identifies the target URL and data to scrape (e.g., tables, paragraphs, prices).
   • Example: A query like “Latest iPhone specs from Apple’s website” returns technical details, pricing, and availability.
3. Response Generation:
   • Cleans and formats scraped data into a concise Slack message.
   • Highlights key details using bullet points or tables (via markdown).
   • Includes warnings if data is unavailable or the site blocks scraping.

Benefits:

1. Automates data collection from websites, saving hours of manual effort.
2. Delivers precise, structured information (e.g., pricing, specs, rankings).
3. Supports research, competitive analysis, or internal reporting.

Add the DumplingAI: Generate Agent Completion module.

• Connection: Use your DumplingAI account.
• Agent ID: 8bfbc60a-36c2-4874-9010-912ed4788e11 (replace with your agent’s ID).
• Messages: Map {{15.text}} (original message text from Slack).
• Parse JSON: false.

Step 2b: Post the Response to Slack

1. Add the Slack: Create Message module.
   • Connection: Use your Slack account.
   • Channel: team-building.
   • Text: Map {{17.text}} (DumplingAI’s response).
   • Use Markdown: true.

Conclusion

This automation enhances Slack communication by integrating Dumpling AI with real-time news search and web scraping. By dynamically processing both text and audio messages, teams can receive automated insights without manual research. Whether tracking breaking news or gathering online information, this workflow keeps your Slack workspace informed and efficient.

Set up this AI-powered Slack automation today and streamline information retrieval like never before!

Get the Blueprint Featured in This Guide

Access the full blueprint here to get started on setting up this automation effortlessly!