How to Build a Slack AI Bot With Bolt for Node

Slack's Bolt framework plus Socket Mode lets you build a bot that responds to mentions without exposing a public endpoint. That makes it perfect for testing from your laptop. Here you will create a Slack app, grab the two tokens it needs, and answer @-mentions with an LLM.

What you need

• A Slack workspace where you can install apps
• Node 18+ installed
• An AI API key
• Permission to create a Slack app at api.slack.com/apps

Step 1: Create the Slack app and enable Socket Mode

At api.slack.com/apps click Create New App, choose From scratch, and pick your workspace. Under Socket Mode, flip it on and generate an app-level token with the connections:write scope. This token starts with xapp-.

Step 2: Add scopes and subscribe to events

Under OAuth & Permissions add the bot scopes app_mentions:read and chat:write. Under Event Subscriptions, turn events on and subscribe the bot to app_mention. Then click Install to Workspace and copy the Bot User OAuth token that starts with xoxb-.

Step 3: Install Bolt and write the app

import "dotenv/config";
import pkg from "@slack/bolt";
import OpenAI from "openai";

const { App } = pkg;
const ai = new OpenAI();

const app = new App({
  token: process.env.SLACK_BOT_TOKEN,
  appToken: process.env.SLACK_APP_TOKEN,
  socketMode: true,
});

app.event("app_mention", async ({ event, say }) => {
  const prompt = event.text.replace(/<@[^>]+>/g, "").trim();

  const res = await ai.chat.completions.create({
    model: "gpt-5-mini",
    messages: [{ role: "user", content: prompt }],
  });

  await say({
    text: res.choices[0].message.content,
    thread_ts: event.ts,
  });
});

await app.start();
console.log("Slack AI bot is running");

Step 4: Run and mention it

Result

Your Slack workspace now has an AI teammate that answers when mentioned, all without a single inbound webhook. To go to production, deploy the same file to a server and keep the process alive with pm2 or a systemd unit.