Telegram Just Made It 2 Taps to Create an AI Agent Bot — Here's How to Use It

Telegram launched Managed Bots on April 15, 2026 — a feature that lets any user create a personalized AI agent bot in two taps. No code. No terminal. No API keys. You tap a link, confirm in a dialog, and you have your own AI-powered bot with its own username, running on the creator's backend. Pavel Durov announced it himself on X, and the AI community immediately started calling it "Lobster Father" — a nod to Telegram's existing BotFather tool and the lobster-themed OpenClaw AI assistant that popularized Telegram as an AI agent platform.

This changes how 950 million Telegram users interact with AI. Here's exactly what Managed Bots are, why they matter, and how to get started — whether you want to use one or build one.

What Are Telegram Managed Bots?

Before this update, creating a Telegram bot meant talking to @BotFather, getting an API token, writing code (or using a no-code platform), and hosting it somewhere. The process took at minimum 30 minutes and required some technical knowledge.

Managed Bots flips this. Now, a developer builds one "manager bot" — the AI backend. Users who want their own personalized version just tap a formatted link, confirm in a pre-filled dialog, and get their own bot instantly. The new bot has its own unique username but runs on the manager bot's backend. The manager receives a webhook notification and gets full API control over the newly created bot.

Think of it like this: the developer builds the engine once, and every user gets their own car with their own license plate — but the same engine under the hood.

This is why people are calling it a potential platform shift. One AI developer can now deploy thousands of individually-branded bots, each personalized to a specific user, without those users touching any code or configuration.

Why "Lobster Father"?

The nickname comes from the collision of two things. First, Telegram's existing bot creation tool is called @BotFather. Second, OpenClaw — the open-source AI agent project by Peter Steinberger — uses a lobster as its mascot and has become the most popular AI agent framework on Telegram. OpenClaw has 247,000 GitHub stars and turned Telegram into the default platform for personal AI agents. When Telegram made it trivially easy to create AI agent bots, the community mashed up "BotFather" with the lobster theme and "Lobster Father" stuck.

It's not an official Telegram name — it's community shorthand for "the new way to create AI bots on Telegram."

How to Use a Managed Bot (As a User)

If someone has built a Managed Bot and shared the creation link, using it takes seconds:

Step 1: Tap the creation link. It looks like t.me/newbot/ManagerBot/YourBotName?name=MyAssistant. The creator typically shares this link on their website, social media, or in a Telegram group.

Step 2: Telegram opens a pre-filled dialog asking you to confirm the bot creation. You'll see the bot's proposed name and username. Tap "Create" to confirm.

Step 3: Your new bot appears in your Telegram chat list. Start messaging it. The bot inherits the AI capabilities of the manager bot but can store your personal preferences, conversation history, and settings.

That's it. Two taps and you have a personal AI assistant running inside Telegram.

How to Build a Managed Bot (As a Developer)

If you want to build the manager bot that others can use to spawn their own bots, here's the technical overview. This requires some development knowledge, but significantly less than building a full bot platform from scratch.

Step 1: Create your manager bot. Start with @BotFather as usual. Create a bot — this will be your "manager" that controls all spawned bots. Save the API token.

Step 2: Set up your backend. Your manager bot needs a server that receives webhook notifications. When a user creates a new bot through your link, Telegram sends a managed_bot webhook update with a ManagedBotUpdated object. Your server calls getManagedBotToken to get the new bot's API token and registers it.

Step 3: Build your AI logic. Connect your manager bot to an AI model — Claude, GPT, Gemini, or a local model. Every message sent to any spawned bot gets routed to your backend, where your AI processes it and sends a response back through that bot's token.

Step 4: Create the spawn link. Format the link as t.me/newbot/YourManagerBot/DefaultUsername?name=DefaultName. Share this link anywhere — your website, social media, Telegram groups.

Step 5: Handle personalization. Each spawned bot has its own token and its own conversation with its user. Store per-user settings, preferences, and conversation history keyed to each bot. This is where the "personalized agent" magic happens.

For a no-code approach, platforms like Botpress, YourGPT, and FastBots already support Telegram bot creation and may add Managed Bot support soon. OpenClaw also integrates natively with Telegram and supports multi-user deployment.

What Can You Build With This?

Personal AI assistants. Each user gets their own bot that remembers their preferences, manages their calendar, answers questions about their documents, or handles specific workflows. OpenClaw already does this — Managed Bots makes it accessible to anyone.

Customer support agents. A business creates one AI support agent, then lets each customer spawn their own dedicated support bot. The customer gets a personal channel for support questions without the business managing thousands of individual bot deployments.

Community bots. A Telegram group admin builds an AI moderator, then shares the creation link. Each sub-community gets their own moderation bot with customized rules and tone, all running on the same backend.

Educational tutors. A course creator builds an AI tutor bot trained on their curriculum. Each student gets their own tutor instance that tracks their progress and adapts to their learning pace.

Trading and finance. Telegram is already massive in crypto communities. A developer builds a portfolio tracker or alert bot, and each user gets a personalized version monitoring their specific holdings.

What to Watch Out For

Security and spam risk. The barrier to creating bots just dropped to zero. Expect a wave of scam bots, phishing attempts, and spam. Only create bots from links shared by sources you trust. Telegram will likely need to add verification and reporting mechanisms as adoption grows.

Data privacy. When you use a Managed Bot, your messages go through the creator's backend. You're trusting the developer with your conversation data. There's no built-in encryption between your spawned bot and the manager's server beyond Telegram's standard encryption. For sensitive conversations, be aware of this.

Backend costs for builders. If you build a popular manager bot and thousands of users spawn their own bots, your AI API costs scale with every user. Plan your pricing or rate limiting accordingly. A bot going viral on Telegram with no cost controls can burn through an API budget fast.

Feature limitations. Managed Bots are new. As of April 2026, you can't yet provide mid-task guidance (the user sends a message and waits for a response — no interactive sessions). Image inputs, file handling, and advanced interactions are supported through the Bot API, but the experience isn't as polished as ChatGPT or Claude's native interfaces.

The Bottom Line

Telegram Managed Bots represents the most frictionless path to giving non-technical users their own AI agent. Two taps, no code, no setup. For builders, it's a distribution channel — build one AI backend, deploy to thousands of users through Telegram's 950 million user base.

The real question isn't whether this will be popular — it will. The question is whether Telegram can handle the trust and safety implications of making AI agent creation this easy. The technology is ready. The guardrails are still being built.

If you're interested in AI agents and coding tools, check out our guides on vibe coding and Claude Code vs Codex, or explore the full State of AI Models comparison to pick the right AI for your project.

Disclosure: Some links in this article are affiliate links. We only recommend tools we've personally tested and use regularly. See our full disclosure policy.