Connecting AI Bots to Slack, Teams and Notion: A Setup Guide to AI Integration by Collaboration Tool

Connecting AI Bots to Slack, Teams and Notion: A Setup Guide to AI Integration by Collaboration Tool

Hand repetitive question handling, meeting-note summaries and data lookups to AI and you can save an average of 47 minutes a day (per the 2024 McKinsey Workplace AI Report). This guide walks you step by step through connecting an AI bot to the three most widely used collaboration tools—Slack, Microsoft Teams and Notion. No coding experience is needed: follow the on-screen instructions and you can finish the setup in under 30 minutes.

What You Need Before You Start

💡 Pro tip: An OpenAI API key is a string that starts with *sk-*. It is shown only once, so copy it right away into a password manager or a notepad.

Part 1: Connect an AI Bot to Slack

Before vs After

Method A: Install the Official Slack ChatGPT App (Easiest)

Step 1. Click the Add apps (+) button at the bottom of the left sidebar in Slack.

Step 2. Type ChatGPT in the search box, then select the “ChatGPT” app made by OpenAI.

Step 3. Click “Add to Slack”, then “Allow.” If an OpenAI account login pop-up appears, log in.

Step 4. Once installation is complete, type the /chatgpt slash command in any channel to check that the bot responds.

Step 5. To make it work only in specific channels: go to Channel settings (⚙️) → Integrations → Add apps → select ChatGPT.

💬 Prompt example 1 — Thread summary:

/chatgpt Summarize this thread in 3 lines and list the action items as bullet points

💬 Prompt example 2 — Customer reply draft:

/chatgpt Write a polite draft reply to this customer complaint: “My delivery was 3 days late. Isn’t there any compensation?”

💡 Pro tip: The Slack ChatGPT app requires a ChatGPT Plus or Enterprise subscription. If you want to use it for free, follow Method B below.

Method B: Build a Custom AI Bot with Zapier (Free Option Available)

This method calls the OpenAI API directly, so you can customize the prompt freely.

Step 1. Go to zapier.com, sign up, and log in.

Step 2. In the left menu, click “Zaps” → “+ Create” at the top right → “New Zap.”

Step 3. Set up the trigger:

• App: Search for and select Slack

• Event: Select New Message Posted to Channel

• Account: Connect your Slack account (click Allow in the OAuth pop-up)

• Channel: Choose the channel where the AI bot will run (e.g., #ai-assistant)

• Click "Continue" → "Test trigger" to confirm the message is picked up

Step 4. Set up the action (AI processing):

• Click the "+" button → search for OpenAI under App and select it

• Event: Select Send Prompt

• Account: Connect by entering your OpenAI API key

• Model: Select gpt-4o (or gpt-4o-mini to save on cost)

• Prompt field: Enter the following:

💬 Prompt example 3 — Setting the system prompt:

You are an internal assistant at a B2B SaaS company. When an employee asks a question in Slack, answer concisely in Korean based on internal company policies and product information. Keep answers to three sentences or fewer, and mark any information you are not sure about with “Needs verification.”

• User Message field: Drag in the Message Text variable from the Slack trigger

Step 5. Send the response back to Slack:

• Click the "+" button → App: Slack → Event: Send Channel Message

• Channel: Select the same channel as the trigger

• Message Text: Insert the Response variable from the OpenAI action

• Bot Name: AI Assistant (or any name you like)

Step 6. Click the "Publish" button at the top right to activate the Zap.

Step 7. Send a test message in that Slack channel:

What is our company's refund policy?

If the AI bot replies within 10–15 seconds, it works.

💡 Pro tip: Zapier’s free plan allows 100 runs per month, which is plenty for teams of 10 or fewer. If you go over, switch to Make.com (formerly Integromat), which offers up to 1,000 free operations.

Part 2: Connecting an AI Bot to Microsoft Teams

Before vs After

Option A: Use Microsoft 365 Copilot (paid license)

If you already have a Microsoft 365 Copilot license ($30/user/month), you can use it directly in Teams with no extra setup.

Step 1. Open Teams and click the "Chat" icon in the left sidebar.

Step 2. Click the Copilot icon (the sparkle ✨) next to the search bar at the top.

Step 3. When the Copilot chat window opens, you can start typing your questions right away:

💬 Prompt example 4 — Meeting summary:

Summarize this morning’s 10 a.m. “Q2 Sales Strategy Meeting” and put each attendee’s action items in a table

💡 Pro tip: Copilot can summarize a meeting only if Teams meeting recording/transcription (Transcript) is enabled. In the admin center, go to Meeting policies → set "Allow transcription" to On.

Option B: Build a free custom bot with Power Automate + OpenAI

Step 1. Log in to make.powerautomate.com with your company Microsoft account.

Step 2. Click "+ Create" → "Automated cloud flow".

Step 3. Enter Teams AI Bot as the flow name.

Step 4. Set up the trigger:

• Type Microsoft Teams in the search box

• Select “When a new channel message is added”

• Team: Choose the team you want (e.g., “Sales Team”)

• Channel: Choose the channel you want (e.g., “AI Assistant”)

Step 5. Add an HTTP action (call the OpenAI API):

• Click “+ New step”, then search for and select “HTTP”

• Method: POST

• URI: https://api.openai.com/v1/chat/completions

• Headers:

  • Content-Type : application/json

  • Authorization : Bearer sk-YourAPIKey

• Body:

json

{

"model": "gpt-4o-mini",

"messages": [

{"role": "system", "content": "You are a sales team AI assistant. Answer questions concisely and practically."},

{"role": "user", "content": "@{triggerBody()?['body']?['content']}"}

],

"max_tokens": 500

}

Step 6. Send the response to Teams:

• “+ New step” → Microsoft Teams → “Post message in a chat or channel”

• Team/Channel: Set the same as the trigger

• Message: In the expression editor, enter body('HTTP')?['choices'][0]?['message']?['content']

Step 7. Click “Save” at the top right, then “Test” → “Manually” → “Run test”.

Step 8. Test it by sending a message in that Teams channel:

Which top 3 industries should we prioritize among this month's new leads?

💡 Pro tip: The HTTP connector isn’t available on the free Power Automate plan. Activate the Premium trial (90 days free), or use Zapier/Make as an alternative. The Zapier setup is the same as Method B in the Slack section—just change the trigger and final action app to Microsoft Teams.

Part 3: Connecting an AI Bot to Notion

Before vs After

Method A: Use Notion AI’s built-in features

Notion has Notion AI built in, so you can use it right away without any integration (extra $10/month per user).

Step 1. Open any Notion page and press the space bar on an empty line, or type /ai.

Step 2. When the AI menu appears, choose the task you want:

• Summarize: Condense a long document into 3–5 lines

• Translate: Convert English materials into Korean

• Brainstorm: Generate a list of ideas

• Custom prompt: Give any instruction you like

Step 3. Select existing text and click the “Ask AI” button (purple ✨) to get an AI response that uses the selected text as context.

💬 Prompt example 5 — Proposal draft:

Based on these client meeting notes, write a draft proposal for a B2B SaaS solution. Include specific figures for the benefits (time savings, ROI), and keep the tone professional yet friendly.

💬 Prompt example 6 — Database analysis:

From the sales pipeline table below, extract the deals in the "Negotiation" stage worth KRW 50 million (≈ US$37,400) or more, and sort them by expected closing probability, highest first

💡 Pro tip: Notion AI uses only the content of the current page as context. To pull in information from multiple pages, turn on Notion AI Q&A (beta): go to Settings → Notion AI → set "AI Q&A" to On.

Method B: Build an Auto-Update Bot with the Notion API + OpenAI

This method sets up a workflow in which AI fills in the content automatically whenever a new item is added to a Notion database. For example, a sales rep enters only a company name, and the AI writes a company summary for them.

Step 1. Create a Notion integration:

• Go to notion.so/my-integrations

• Click "+ New integration"

• Name: AI Content Bot

• Select your workspace → click "Submit"

• Copy and save the generated Internal Integration Secret (it starts with ntn_)

Step 2. Connect the integration to your database:

• Open the Notion database page you want to connect to the AI bot

• Click the "···" menu at the top right → "Add connections" → select "AI Content Bot" → "Confirm"

Step 3. Build the workflow in Zapier:

• In Zapier, click "+ Create" → "New Zap"

Trigger:

• App: Notion → Event: New Database Item

• Select your database (e.g., "Lead Management DB")

Action 1 — AI processing:

• App: OpenAI → Event: Send Prompt

• Prompt:

💬 Prompt example 7 — Automated company research:

Summarize information useful for B2B sales about the following company: {{Company name}}. Include: 1) Business areas 2) Recent news 3) Likely pain points 4) How it connects to our solution. Keep each item to two sentences or fewer.

• In the {{Company name}} spot, insert the variable for the "Company name" property from the Notion trigger.

Action 2 — Notion update:

• App: Notion → Event: Update Database Item

• Database Item: select the Item ID from the trigger

• Insert the OpenAI response into the "AI Research" property (Rich Text)

Step 4. Click "Publish", then test it by adding a new company to your Notion database.

💡 Pro tip: Add an "AI Status" select property to your Notion database (Pending → Done), and have the last Zapier step set it to "Done". You can then see at a glance which items the AI has processed.

Cross-Platform Automation Connecting Three Tools

The most powerful approach is to chain Slack → AI → Notion → Teams into a single workflow.

Real-World Scenario: Sales Meeting Follow-Up Automation

1. A sales rep writes meeting notes in the #meeting-notes channel in Slack

2. AI (OpenAI) analyzes the meeting content and generates a summary plus a draft follow-up email

3. Notion's sales pipeline DB automatically gets a new item, with the AI analysis recorded in it

4. A "New meeting completed" notification is sent to the sales manager channel in Teams

Build this workflow in Zapier as a Multi-Step Zap, connecting the triggers and actions for each tool covered above in order.

Expected Results (for a 10-person sales team)

Troubleshooting & FAQ

Checklist for When the Bot Doesn’t Respond

• Is your API key valid? Check your credit balance in the OpenAI dashboard.

• Are the app permissions correct? Make sure the bot has been added to the channel in Slack/Teams.

• Is Zapier/Power Automate active? Check that the Zap/flow is set to “On” in the dashboard.

• Is the Notion integration connected to the page? Check the page’s “···” menu → “Connections.”

Cost Management Tips

• Using the gpt-4o-mini model cuts costs by about 90% compared with gpt-4o (and its quality is enough for most work questions).

• Capping max_tokens at 500 prevents needlessly long responses.

• Add a filter step in Zapier so only messages containing a specific keyword (e.g., @ai or #question) are passed to the AI, and you can cut API calls by 60–70%.

💡 Pro tip: If you’re worried about monthly API costs, go to the OpenAI dashboard → Usage limits and set a monthly cap of $10. Once the limit is reached, API calls are blocked automatically, so you avoid unexpected charges.

Next steps: enhancement roadmap

Completing just Phase 1 of this guide can bring an immediate, noticeable efficiency gain to your team’s daily work. Start with a single tool today—connecting an AI bot to the one channel you use most takes only 30 minutes.