If you’ve ever wished your AI could live where you already chat—WhatsApp, Telegram, Slack, Discord—while staying under your control, Clawdbot is built for exactly that. Clawdbot (often stylized as ClawdBot) is an open-source, self-hosted personal AI assistant you run on your own device or server, and then talk to through your favorite messaging apps. (github.com)
Unlike a normal chatbot that only responds when you open a website, Clawdbot is designed to feel local, always-on, and practical: it’s driven by a “Gateway” that connects your chat apps to an AI agent, plus a local Control UI (dashboard) for configuration and monitoring. (github.com)
What is Clawdbot?
Clawdbot is a personal AI assistant you host yourself. Its core idea is simple: you message it from the apps you already use, and it replies there—so your “AI assistant” feels like a normal contact in your chat list, not another tab you have to remember to open. The project describes itself as a personal assistant that can answer on WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage, Microsoft Teams, WebChat, and more. (github.com)
Clawdbot is also “local-first” in the sense that the main control plane (the Gateway) runs on your machine. The docs describe it as a bridge that connects chat channels (like WhatsApp/Telegram/Discord/iMessage) to AI agents (such as Pi) and related tools. (docs.clawd.bot)
In one line
Clawdbot is an open-source AI agent that runs on your device and lets you talk to your assistant through WhatsApp/Telegram/Slack/Discord—so the assistant can be always available and integrated with your daily workflow. (github.com)
Why people are talking about Clawdbot (and why it feels different)
Clawdbot became a hot topic because it tries to solve a real frustration: many AI tools are powerful, but they’re not present in your daily life. According to a recent report, Clawdbot’s “always-on” approach—running locally and connecting to messaging apps—has encouraged some people to use old laptops or Mac minis as a dedicated “agent box,” while also raising important security conversations. (businessinsider.com)
What stands out is the combination of:
- Self-hosting: you run it on your own hardware. (github.com)
- Real chat surfaces: it lives in WhatsApp/Telegram/Discord, not just a web UI. (github.com)
- A control plane (Gateway + dashboard): there’s an explicit architecture for routing, sessions, and configuration rather than “just a chat window.” (docs.clawd.bot)
How Clawdbot works (Gateway, chat channels, agents, and sessions)
To understand Clawdbot, picture it like a switchboard:
- You send a message from WhatsApp/Telegram/Discord/iMessage (and other supported channels).
- That message goes to the Gateway, the central component that owns the channel connections and handles routing.
- The Gateway forwards the request to an AI agent (the docs mention connecting to coding agents like “Pi”).
- The agent responds, and the Gateway sends the reply back to the same channel.
The official docs even show this as an architecture diagram and note practical design points like “one Gateway per host (recommended)” and “loopback-first” defaults for security-minded setups. (docs.clawd.bot)
The dashboard (Control UI)
Clawdbot includes a browser-based dashboard/Control UI for chat, configuration, sessions, and more. By default, it’s available locally (for example on http://127.0.0.1:18789/ on the gateway host). (docs.clawd.bot)
This matters because it means you’re not just “chatting with an AI”—you’re running a small system with routing, sessions, and tools, and you can manage it like software.
Read Also : What is NPTEL SWAYAM? Complete Guide to India’s Premier Online Learning Platform
What can you do with Clawdbot? Practical use cases (beginner-friendly)
If you’re new to AI agents, here are realistic ways people think about using Clawdbot:
A personal productivity assistant inside WhatsApp/Telegram
Because Clawdbot is designed to respond through chat apps, it can become your daily command line—without opening a laptop:
- Summarize what you missed (messages, notes, updates)
- Draft replies and save time on repetitive communication
- Remind you about tasks and meetings
The project’s website positions it as an AI that can manage inbox and calendar-style tasks from chat. (clawd.bot)
A developer helper that feels always-on
The docs explicitly describe Clawdbot as a gateway for AI agents (including coding agents). That makes it interesting for developers who want an “assistant running near the code,” reachable from messaging apps. (docs.clawd.bot)
A single assistant across many platforms
Many people already coordinate life across multiple apps—WhatsApp for family, Telegram for groups, Slack for work, Discord for communities. Clawdbot’s multi-channel approach lets you keep one assistant identity while separating sessions appropriately (for example, DMs vs groups). (github.com)
A private-ish “home base” for your AI
Because you run the Gateway yourself, you control where configuration and logs live. It’s still important to understand that the model you choose may be cloud-hosted (OpenAI/Anthropic) unless you run a local provider—but the overall architecture is more controllable than “everything happens on someone else’s website.” (github.com)
Which chat apps does Clawdbot support?
From the project’s own description, Clawdbot can answer on common channels such as:
- Telegram
- Slack
- Discord
- Google Chat
- Signal
- iMessage
- Microsoft Teams
- WebChat
…and it also mentions extension channels like BlueBubbles, Matrix, and Zalo variants. (github.com)
If you’re a beginner, the key takeaway is: it meets you where you already communicate—you don’t need to learn a new app just to talk to your assistant.
What AI models can Clawdbot use?
Clawdbot doesn’t ship with “one fixed brain.” Instead, you connect it to an AI model provider and choose a model.
Cloud models (Claude, OpenAI, and others)
The GitHub project notes support for subscriptions and authentication flows with providers like Anthropic and OpenAI. (github.com)
Local models (Ollama)
If you prefer to run models on your own machine, Clawdbot can integrate with Ollama, a local LLM runtime. The docs explain that Clawdbot can connect via Ollama’s OpenAI-compatible API and even auto-discover tool-capable models when configured appropriately. (docs.molt.bot)
This is an important option for people who care about privacy, latency, or cost control—because local models can reduce dependence on paid cloud APIs (though hardware capability still matters).
Is Clawdbot safe? What you should know before using it
Any tool that connects AI to your real communication channels and potentially to actions on your machine must be treated seriously.
Clawdbot’s own security documentation emphasizes a “trust hierarchy” mindset (who should be trusted vs not) and encourages responsible disclosure for vulnerabilities. (docs.clawd.bot)
Here are practical safety principles (especially if you’re new):
Treat inbound messages as untrusted input
When your assistant lives in chat apps, anyone who can message it can try to manipulate it. This is why beginners should start with direct messages only and avoid adding the bot to large group chats until they fully understand permissions and pairing controls. (The project’s broader security guidance repeatedly points users toward careful trust boundaries.) (docs.clawd.bot)
Keep the Gateway locked down
The docs describe loopback-first defaults (local-only) and discuss binding options for broader access. As a beginner, you should prefer a local-only setup unless you know exactly why you need remote access. (docs.clawd.bot)
Use the least-privilege approach
Even if Clawdbot can be connected to powerful tools, you don’t have to enable everything. Start with simple skills (summaries, drafts, reminders), and expand only when you’re confident.
Understand your model choice
If you use a cloud provider, your prompts and outputs are processed by that provider under their terms. If you use a local model provider like Ollama, you’re shifting more work onto your own machine—and gaining more control. (docs.molt.bot)
How to get started with Clawdbot (high-level steps)
You don’t need to be an expert to understand the setup flow, even if you decide to install later.
Step 1: Check prerequisites
The “Getting started” guide lists Node.js (version 22+) as a prerequisite and strongly recommends WSL2 for Windows users. (docs.clawd.bot)
Step 2: Install the CLI
The docs show a recommended installer script and also mention alternatives like global npm install. (docs.clawd.bot)
Step 3: Run the onboarding wizard
Clawdbot recommends using its onboarding wizard (clawdbot onboard) because it sets up gateway settings, model/auth, channels, pairing defaults, workspace, and (optionally) a background service. (docs.clawd.bot)
Step 4: Start the Gateway and open the dashboard
Once the Gateway is running, you can open the local Control UI (commonly at http://127.0.0.1:18789/) and begin chatting, even before connecting external channels. (docs.clawd.bot)
Step 5: Connect one chat surface (Telegram/WhatsApp/etc.)
After onboarding, you connect the channel you want. For example, Telegram uses a Bot API channel owned by the Gateway, with deterministic routing (replies go back to Telegram). (docs.clawd.bot)
How much does Clawdbot cost?
Clawdbot is open-source and uses the MIT license, so the software itself is free to use and modify. (github.com)
Your real costs usually come from:
- Hardware (your own laptop/server, or a small VPS)
- Model usage (if you use paid cloud models)
If you run a local provider like Ollama, you can reduce or eliminate cloud model fees, but you’ll need sufficient CPU/GPU resources depending on the model you choose. (docs.molt.bot)
Who is Clawdbot for?
Clawdbot is best for people who:
- Want a personal assistant inside WhatsApp/Telegram/Slack instead of a separate AI app (github.com)
- Prefer self-hosted tools and more control over configuration and data (clawd.bot)
- Are comfortable learning a little setup (CLI, onboarding wizard, tokens/keys) (docs.clawd.bot)
It may not be ideal if you want a one-click, fully managed consumer assistant with strict guardrails. The more power you give an agent, the more careful you need to be.
FAQs
Is Clawdbot the same as a normal Telegram bot?
Not really. A typical Telegram bot is often “just chat.” Clawdbot is a system that can connect multiple chat platforms through a Gateway and route requests to AI agents, with a dashboard and configuration model behind it. (docs.clawd.bot)
Does Clawdbot run locally?
Yes—Clawdbot is designed to run on your own devices, with a local Gateway and Control UI. (github.com)
Can I use Clawdbot with a local AI model?
Yes. The official docs describe an Ollama provider integration so you can use locally hosted models through an OpenAI-compatible API. (docs.molt.bot)
What is the “Gateway” in Clawdbot?
The Gateway is the control plane that owns your channel connections and handles routing, sessions, and the dashboard/Control UI. It’s the core service that makes “chat app → AI agent → chat app” work reliably. (docs.clawd.bot)
At the End
Clawdbot is one of the clearest examples of a modern “AI agent” approach: not just a chatbot, but a self-hosted assistant that lives inside real messaging apps and can be configured like software. It combines a local Gateway, multi-channel chat routing, and flexible model providers (including local options like Ollama). (github.com)
If you’re new to Clawdbot, the smartest way to think about it is: a personal AI you control, reachable from your pocket, but powerful enough that you should set it up thoughtfully. (docs.clawd.bot)