Bolt.new AI and Microsoft Teams integration: Step-by-Step Guide 2025

The direct answer: You don’t “integrate Bolt.new AI into Microsoft Teams” directly. Bolt.new is a browser-based workspace, not a Teams app. The correct pattern is: you build a backend (inside bolt.new) that exposes a public HTTPS endpoint, register a Microsoft Teams bot in Azure Bot Framework, configure its messaging endpoint to point to your Bolt backend, handle Teams events via Microsoft Bot Framework REST schemas, and optionally wrap your Bolt-based logic behind a Teams message extension or slash-command–style bot. Teams talks to your server via standard Bot Framework webhooks; your server (running in bolt.new) talks back using the Bot Framework API.

Below is the clear, real, step‑by‑step explanation of how to make a Bot Framework–based Teams integration work using a backend built inside bolt.new.

 

What We Are Actually Integrating

 

Microsoft Teams does not connect directly to an AI workspace. It only talks to:

• Azure Bot Framework (via a registered bot)
• Webhooks that match the Bot Framework protocol
• Message Extensions (still Bot Framework)
• Optional OAuth flows for user identity

Bolt.new acts as your development sandbox where you host the backend code, usually an Express.js server, that receives Teams events and sends replies.

 

High-Level Flow

 

• You create an Express API in bolt.new.
• You expose it publicly using Bolt’s temporary public URL.
• You register a bot in Azure Bot Framework and set its messaging endpoint to your Bolt public URL.
• Teams messages (e.g., “Hello bot”) trigger POST requests to your Bolt server.
• Your Bolt server processes the message (possibly using an AI model/API).
• Your server responds using the Bot Framework’s reply schema.

 

Step‑by‑Step: Integration That Actually Works

 

This is the cleanest, simplest, and fully real-world approach.

• Create a Bot in Azure Portal In Azure Portal, search for “Azure Bot”. Create a new bot resource. You will get: MicrosoftAppId, MicrosoftAppPassword (client secret).
• Enable Teams channel Inside the bot resource → Channels → Add Microsoft Teams.
• Set the Bot Messaging Endpoint Teams expects something like: https://your‑bolt‑public‑url/api/messages In bolt.new this is the temporary public URL you get when you “share” the project.
• Create the Bolt backend Inside bolt.new, create a simple Express server that can handle Bot Framework messages. Use the official botbuilder npm package.
• Store credentials Put MicrosoftAppId and MicrosoftAppPassword in bolt.new environment variables.

 

Minimal Working Example (Express + BotBuilder inside bolt.new)

 

import express from "express";
import { BotFrameworkAdapter } from "botbuilder";

const app = express();

// Load secrets from environment variables
const adapter = new BotFrameworkAdapter({
  appId: process.env.MICROSOFT_APP_ID,        // Provided by Azure Bot registration
  appPassword: process.env.MICROSOFT_APP_PASSWORD
});

// Basic bot logic
adapter.onTurn(async (context) => {
  if (context.activity.type === "message") {
    await context.sendActivity("Hello from Bolt.new backend! You said: " + context.activity.text);
  }
});

// Teams will POST here
app.post("/api/messages", (req, res) => {
  adapter.processActivity(req, res, async (context) => {
    // Bot logic handled above
  });
});

// Start server
app.listen(3000, () => {
  console.log("Bolt.new bot server running on port 3000");
});

 

This code is fully valid and matches Microsoft Bot Framework SDK for Node.js.

 

Testing the Integration

 

• Run the server inside bolt.new.
• Expose the public URL.
• Copy the public URL → paste into Azure Bot “Messaging endpoint”.
• Open Teams → Add the bot using its App ID.
• Send a message → “Hello”.
• Your Bolt server receives the webhook → responds → message appears in Teams.

 

If You Want AI Responses

 

Inside the if (context.activity.type === "message") block, call an LLM API such as OpenAI or Azure OpenAI. Example:

import OpenAI from "openai";

const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

// ...
if (context.activity.type === "message") {
  const userText = context.activity.text;

  const ai = await client.chat.completions.create({
    model: "gpt-4o-mini",
    messages: [{ role: "user", content: userText }]
  });

  await context.sendActivity(ai.choices[0].message.content);
}

 

Deployment Outside Bolt.new

 

• Deploy the same Express bot to Azure App Service, Vercel, Render, or any server.
• Update the bot messaging endpoint to the permanent domain.
• Keep secrets in environment variables.

This approach is exactly how production-grade Teams bots are built.