How to Build an AI Telegram Bot with Claude API and Cloudflare Workers

AI Chat Assistants on the Edge

Building a chat assistant directly inside messaging platforms (like Telegram or WhatsApp) is a highly effective way to automate customer support or build consumer AI utilities.

By leveraging Cloudflare Workers (serverless runtime) and Anthropic's Claude API, you can deploy a scalable AI assistant that processes webhooks instantly, runs on cheap credits, and responds within seconds.

This tutorial walks through creating a Telegram bot that consumes webhooks and answers using Claude.

1. Setting Up Your Telegram Bot Token

1. Search for @BotFather on Telegram.

2. Send `/newbot` and follow the instructions to set a name and username.

3. Save the generated API HTTP Token (e.g., `7283921829:AAHg82...`).

2. Project Initializer

Create a TypeScript worker with Hono:

npm create cloudflare@latest ai-bot -- --template=hono
cd ai-bot
npm install

Install the official Anthropic SDK client:

npm install @anthropic-ai/sdk

3. Webhook Logic & Claude Call

Update `src/index.ts` to parse incoming Telegram messages and fetch replies from Claude:

import { Hono } from 'hono';
import Anthropic from '@anthropic-ai/sdk';

type Bindings = {
  TELEGRAM_TOKEN: string;
  ANTHROPIC_API_KEY: string;
};

const app = new Hono<{ Bindings: Bindings }>();

app.post('/webhook', async (c) => {
  const body = await c.req.json();
  const message = body?.message;

  if (!message || !message.text) {
    return c.text('OK'); // Ignore non-text messages
  }

  const chatId = message.chat.id;
  const userText = message.text;

  // Initialize Claude Client
  const anthropic = new Anthropic({
    apiKey: c.env.ANTHROPIC_API_KEY,
  });

  // Call Claude
  const response = await anthropic.messages.create({
    model: 'claude-3-5-sonnet-latest',
    max_tokens: 1000,
    messages: [{ role: 'user', content: userText }],
  });

  const replyText = response.content[0].text;

  // Send message back to Telegram
  const telegramUrl = `https://api.telegram.org/bot${c.env.TELEGRAM_TOKEN}/sendMessage`;
  await fetch(telegramUrl, {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      chat_id: chatId,
      text: replyText,
    }),
  });

  return c.text('OK');
});

export default app;

4. Deploying & Registering the Webhook

Encrypt your secrets in wrangler:

npx wrangler secret put TELEGRAM_TOKEN
npx wrangler secret put ANTHROPIC_API_KEY
npx wrangler deploy

Register your deployed worker URL as the official Telegram webhook receiver:

curl -X POST "https://api.telegram.org/bot<YOUR_TELEGRAM_TOKEN>/setWebhook"      -H "Content-Type: application/json"      -d '{"url": "https://your-worker.your-subdomain.workers.dev/webhook"}'

Your bot is now live! Send a message to your Telegram bot, and Claude will respond.

CTA: AI App Consulting

I built ReplyGeniusAI, an AI assistant handling context-aware email and chat routing with smart prompt pipelines. Let's build your AI product: email me at [hello@anmolmaan.dev](mailto:hello@anmolmaan.dev).