Most AI agents force you to use a clunky web dashboard or a boring terminal window. It’s annoying because that’s not where you actually spend your time. You’re on Slack for work, Telegram for friends, and maybe WhatsApp for family. Why should your AI live somewhere else? OpenClaw is built to meet you where you already are. You can connect it to over twenty different platforms with just a few lines of configuration. Whether you want a research assistant in Discord or a productivity bot in Telegram, this guide will show you how to set it up in minutes.
How do platform connections actually work?
OpenClaw treats every messaging app the same way. You get a token or a set of credentials from the platform, drop them into your config.yaml, and the agent starts listening. You don’t have to write any platform-specific code. The agent just sees “Message Received” and sends a “Message Reply.” It’s a clean abstraction that makes it easy to swap apps or run on multiple platforms at once without complicating your logic.
The Scenario: You’re switching from a startup that uses Slack to a bigger company that uses Discord. Instead of rebuilding your research agent, you just add a Discord token to your config. The agent keeps all its memory and skills, but now it talks to you in a different app.
How do I set up a Telegram bot?
Telegram is by far the easiest integration to set up. You talk to @BotFather, get a token, and you’re basically done. It’s reliable, fast, and supports Markdown formatting. Make sure you use the allowedUsers field to restrict the bot to your own ID. If you don’t, anyone who finds your bot’s username can start chatting and burning through your AI API credits.
messaging:
- platform: telegram
botToken: "your_token_here"
allowedUsers: ["123456789"] # Use your numeric IDSearch for @userinfobot in Telegram to find your numeric ID if you don’t know it. Don’t use your username; it has to be the ID.
The Scenario: You’re on a long flight with terrible Wi-Fi. Telegram’s low-data mode is the only thing that works. You can still message your agent to summarize some notes because the integration is lightweight and handles spotty connections perfectly.
Can I use OpenClaw with WhatsApp?
Yes, but it’s a bit different. Since WhatsApp doesn’t have an official bot API for personal accounts, OpenClaw uses a library that mimics a web session. You scan a QR code with your phone just like you’re linking a new laptop. Once it’s linked, you can message yourself (the “Saved Messages” chat) and the agent will reply. It’s great for when you’re on the move and don’t want to switch apps.
The Scenario: Your family group chat is blowing up with a hundred messages while you’re at work. You forward a few to your “Saved Messages” chat and ask your agent to “Summarize the drama.” It gives you the three bullet points that actually matter, and you can get back to your day.
How do I connect OpenClaw to Slack?
Slack is a bit more involved because you have to create a “Slack App” and enable “Socket Mode.” You’ll need two tokens: a Bot Token (starts with xoxb) and an App Token (starts with xapp). It’s a bit of a click-fest in the Slack API dashboard, but it’s worth it for work-related tasks. You can even restrict the agent to specific channels if you don’t want it eavesdropping on everything.
- Enable
chat:writeandim:historyscopes. - Enable Socket Mode for real-time responses.
- Add your Slack member ID to the allowed list.
The Scenario: You’re in the middle of a busy sprint and need to check a Jira ticket status. Instead of opening a new tab, you mention your agent in a private Slack channel. It pulls the data and replies in the thread, keeping you in the flow of your work conversation.
What about Discord integration?
Discord is similar to Telegram—you create a bot in the developer portal, enable “Message Content Intent,” and invite it to your server. It’s popular for developer-focused agents because you can use code blocks and mentions. You can also set allowedChannels so the agent doesn’t reply to every single random comment in a general chat.
messaging:
- platform: discord
botToken: "your_token"
allowedUsers: ["your_username"]
allowedChannels: ["agent-chat"]It’s great for building “team” agents that everyone in a small server can use for specific tasks like code reviews or documentation lookups.
The Scenario: You and your friends are debating the best way to deploy a React app. You mention the agent in your
#dev-chat. It searches the web, provides a cited answer, and even formats the code for you. No one had to leave the chat to win the argument.
How does voice mode work on mobile?
If you’re using the OpenClaw companion app on iOS or Android, you can use voice mode. It’s not just “speech-to-text”; it’s a full conversational interface. You can set a silenceTimeoutMs in your config so the agent waits for you to finish your thought before it starts processing. This makes the conversation feel much more natural, like you’re actually talking to someone rather than barking commands at a machine.
The Scenario: You’re driving and have a great idea for a blog post. You tap the voice button on the OpenClaw app and dictate your thoughts. The agent cleans up your ramblings, identifies the key themes, and saves it all to your “Ideas” folder for later. No typing required.
What to Read Next
- Secure your data: How OpenClaw Memory and Privacy Work
- Build your first agent: OpenClaw Tutorial: Your First AI Agent
- Install guide: How to Install OpenClaw on Ubuntu, macOS, and Windows