Clawdbot
Open Source | AI Agents
Overview
OpenClaw is an open-source personal AI assistant that lives on your own Mac, Windows, or Linux machine, not in a walled-garden cloud service. Unlike SaaS AI tools, your data and context stay entirely on your hardware. You connect it to the messaging apps you already use, including WhatsApp, Telegram, Discord, Slack, Signal, and iMessage, and interact with it through direct messages or group chats. OpenClaw uses your own AI model subscriptions, supporting Anthropic Claude, OpenAI GPT, and locally running models, so you control both the compute costs and the data privacy. The tool ships with persistent memory, meaning it learns your preferences and context over time and carries that knowledge across every conversation. It can take proactive actions on a schedule via cron-style jobs, send reminders, manage your calendar, check you in for flights, and handle email. A growing community publishes shareable skills and plugins, and the tool can write and install new skills on its own. A companion menubar app for macOS (beta, requires macOS 15 or later) complements the command-line installer.
Features
- Multi-platform messaging: Works through WhatsApp, Telegram, Discord, Signal, Slack, and iMessage simultaneously
- Persistent memory: Remembers all conversations and preferences stored locally as Markdown files
- Proactive messaging: Sends morning briefings, calendar alerts, and reminders without being asked
- Email management: Sorts, summarizes, and drafts responses to your emails automatically
- Calendar scheduling: Books meetings, sends invites, and handles rescheduling through natural language
- Flight check-in: Automatically checks you in for flights 24 hours before departure
- Browser automation: Fills forms, extracts data, and navigates websites hands-free
- Local file and system access: Reads, writes files, and executes shell commands on your machine
- Privacy-first architecture: All data stored locally; nothing leaves your machine except API calls
- Open-source (MIT license): Full source code available, free to use, modify, and extend
- Skills and plugins: Extend with community-built skills or create your own custom automations
- Model flexibility: Works with Claude, GPT, or other models via API; supports failover between providers
- Smart home control: Manages smart devices and routines through natural conversation
- Research and summaries: Browses the web and delivers concise summaries on any topic you request
Best For
Technical users who want a privacy-first AI assistant running entirely on their own hardware, Power users managing busy schedules across email, calendar, flights, and multiple chat platforms, Developers and tinkerers who want an extensible, open-source assistant they can customize with plugins, Teams and families sharing a self-hosted assistant for coordinating tasks, reminders, and logistics
How It Works
OpenClaw installs via a single npm command or a one-line curl script that also sets up Node.js and all dependencies automatically. After installation, you run an onboarding command that connects your chosen messaging platforms and AI model API keys. From that point, you talk to OpenClaw through any connected chat app exactly as you would talk to a person. When you send a message, OpenClaw routes it to your chosen AI model, executes any required actions on your machine such as reading or writing files, running shell commands, opening a browser, or calling third-party integrations like Gmail, GitHub, Spotify, or Obsidian, and then sends the result back to you in chat. Persistent memory means context carries forward across sessions. Background heartbeat tasks let it perform scheduled jobs without any input from you. Skills extend its capabilities, and because the source code is open, you can fork and modify any behavior yourself.