Quevin

Two Claudes, One Telegram: Running Claude Code and OpenClaw Side by Side

Kevin P. Davison
AI Claude OpenClaw Claude Code Telegram Developer Tools
Two Claudes, One Telegram: Running Claude Code and OpenClaw Side by Side

Two Claudes, One Telegram: Running Claude Code and OpenClaw Side by Side

I now have two separate Claude-powered bots on Telegram. They look similar from the outside — both respond to messages, both have access to my codebase, both run on Claude Sonnet — but they’re architecturally different tools built for different jobs.

Here’s how I set them up and why I run both.


Claude Code on Telegram

Anthropic recently published an official Telegram plugin for Claude Code. The setup is clean:

  1. Create a bot via BotFather, grab the token
  2. Install the plugin inside a Claude Code session: /plugin install telegram@claude-plugins-official
  3. Configure it with your token: /telegram:configure <token>
  4. Relaunch with the channel flag: claude --channels plugin:telegram@claude-plugins-official
  5. DM your bot — it replies with a 6-character pairing code
  6. Pair it inside Claude Code: /telegram:access pair <code>

That’s it. Your next DM goes directly into a live Claude Code session. The bot can read files, write code, run shell commands — everything Claude Code can do — delivered over Telegram.

The key thing to understand: Claude Code on Telegram is a code execution environment. It’s Claude Code with a chat interface bolted on. It lives in a terminal session, it has your full development environment, and it’s particularly good at iterative coding tasks: write a function, test it, fix it, commit it. When you close the terminal, the session ends.


OpenClaw on Telegram

OpenClaw is a different animal. Where Claude Code on Telegram is a coding terminal with a chat UI, OpenClaw is a persistent personal agent with a tool ecosystem.

Setup was equally straightforward — create a bot via BotFather, drop the token in ~/.openclaw/openclaw.json, and the gateway handles the rest. But what connects is a long-running agent with access to a skill library: email, calendar, Apple Reminders, iMessage via BlueBubbles, image generation, GitHub, web search, and more.

The session persists between messages. It has memory across restarts. It can initiate actions proactively via heartbeats — checking email, syncing files, watching for events. It’s less “terminal with chat” and more “assistant that happens to be in your terminal.”


The Practical Difference

Claude Code (Telegram plugin)OpenClaw (Telegram)
Primary jobWrite, run, and fix codeOrchestrate tools and tasks
Session modelLives in a terminal sessionPersistent, survives restarts
MemoryContext window onlyFile-based memory across sessions
Tool accessFull dev environment (fs, shell, git)Skill ecosystem (email, calendar, iMessage, etc.)
Best forCoding tasks, PR reviews, refactorsDaily workflows, automation, communications
Initiated byYou start a session, then chatAlways running, can reach out proactively

In practice, I use them for different things. When I want to iterate on code — write a component, fix a bug, review a PR — I reach for Claude Code. When I want to get something done that touches multiple systems — check my calendar, send a message, post a blog, manage reminders — I reach for OpenClaw.


Why Both?

The honest answer: I didn’t plan it this way. I built OpenClaw first because I wanted a persistent agent. Claude Code’s Telegram plugin came later and solved a different problem — I wanted to do coding work from my phone without opening a laptop.

They’ve ended up complementary. OpenClaw is the agent. Claude Code is the power tool. Both happen to live in Telegram, which means I have a surprisingly capable development environment and personal assistant accessible from anywhere, through the same app I use to text people.

That’s the setup. More on how I use both in practice as the experiment continues.


Kevin is a Senior Technical Lead at Fluke Corporation.

Kevin P. Davison

About the Author

Kevin P. Davison has over 20 years of experience building websites and figuring out how to make large-scale web projects actually work. He writes about technology, AI, leadership lessons learned the hard way, and whatever else catches his attention—travel stories, weekend adventures in the Pacific Northwest like snorkeling in Puget Sound, or the occasional rabbit hole he couldn't resist.