Frequently Asked Questions

OpenClaw is an open-source personal AI assistant that connects to multiple messaging platforms like WhatsApp, Telegram, Discord, and Slack. It allows you to run your own AI assistant with full privacy and control.

Yes, OpenClaw is completely free and open source under the MIT license. You only need to pay for the AI model API you choose (e.g., Anthropic Claude or OpenAI GPT). Many models offer free tiers.

Install OpenClaw using npm: npm install -g openclaw. Then run openclaw onboard to configure your first channel. For detailed instructions, see our Installation Guide.

OpenClaw supports 20+ AI models including Anthropic Claude (Opus 4, Sonnet 4), OpenAI GPT-5.1 Codex, GPT-5.3 Codex, Google Gemini 3, GLM/Z.AI, DeepSeek, Kimi, Mistral, Llama, Qwen, and more. You can switch models by changing one line in your configuration.

OpenClaw is designed with security in mind. As a self-hosted solution, your data never leaves your server. We recommend following our Security Hardening Checklist for production deployments.

Yes, OpenClaw supports WhatsApp integration via Baileys. You can connect your AI assistant to WhatsApp in minutes using QR code pairing.

OpenClaw is built with TypeScript and Node.js, making it easy for developers to extend and customize.

Unlike ChatGPT (cloud-based), OpenClaw is self-hosted, giving you complete privacy and data control. OpenClaw also supports multi-channel integration (WhatsApp, Telegram, Discord) which ChatGPT does not.

Yes, OpenClaw is designed to be self-hosted. You can run it on any server with Node.js 22+, including cloud VPS, home servers, or even a Raspberry Pi.

Yes, OpenClaw supports voice messages on channels that support it (WhatsApp, Telegram, Discord). You can use voice input and receive voice responses.

OpenClaw has a Skills system that allows you to extend its capabilities. Check our Skills documentation for guides on creating custom skills.

OpenClaw runs on macOS, Linux, and Windows (via WSL2). Companion apps are available for iOS and Android.

Yes, OpenClaw supports Chinese AI models including GLM (Z.AI), DeepSeek, Kimi (Moonshot), Qwen, and Doubao (Volcano Engine). These can be configured through custom providers.

BlueBubbles is the recommended method for iMessage integration. It provides full feature support including edit, unsend, reactions, and group management. Legacy iMessage via imsg CLI is deprecated.

OpenClaw requires Node.js 22+ and runs on macOS, Linux, or Windows (via WSL2). Minimum 2GB RAM recommended. For production, 4GB+ RAM and a VPS with static IP is recommended.

Basic setup takes 5-10 minutes. Install with npm, run openclaw onboard to configure your AI model and channel, then start using your AI assistant. Full configuration with all features may take 30-60 minutes.

OpenClaw requires an AI model API key to function. You can use free tiers from providers like OpenAI (GPT-4o free), Anthropic (Claude free), or Google Gemini. For local models, you can use Ollama with no API cost.

Run npm install -g openclaw@latest to update. Check your current version with openclaw --version. Always backup your configuration before updating.

You can connect unlimited channels to OpenClaw. However, each channel requires its own authentication (Bot Token, QR pairing, etc.). Most users connect 2-5 channels for different use cases.

Yes, OpenClaw works in group chats on most platforms (WhatsApp, Telegram, Discord, Slack). You can configure mention keywords, auto-respond rules, and privacy settings for group interactions.

Yes, one OpenClaw instance can manage multiple channels simultaneously. All channels share the same AI model, memory, and configuration. You can customize per-channel responses if needed.

Open BotFather on Telegram (@BotFather), use /newbot command, follow prompts to name your bot, then copy the API token. Never share this token publicly.

Common issues: (1) Session expired - reconnect with QR code, (2) Network blocked - use VPN or proxy, (3) Phone disconnected - ensure phone has stable internet. Check logs with openclaw connect whatsapp --debug.

OpenClaw is self-hosted, meaning your data never leaves your server. All conversations stay local. Enable encryption at rest, use 2FA, and follow our Security Hardening Guide for production deployments.

OpenClaw processes messages locally on your server to generate AI responses. No messages are sent to third-party servers unless you explicitly configure cloud logging. You have full control over data retention policies.

Configure 2FA through your dashboard settings. Use authenticator apps (Google Authenticator, Authy) or hardware keys (YubiKey). Enable 2FA for both the web UI and SSH access to your server.

Conversation logging is disabled by default. You can enable it in settings for debugging or context recall. All logs stay on your server and can be automatically deleted after N days.

For general use: Claude 4.5 Sonnet offers best balance of capability and cost. For coding: GPT-5.1 Codex excels. For free: GPT-4o or Claude 3.5 Sonnet free tiers. For privacy: use Ollama with Llama 3.3 locally.

OpenClaw itself is free. Costs depend on your AI model: GPT-4o ($0-10/month for typical use), Claude 4.5 Sonnet ($5-20/month), Gemini 3 (often free). Server costs: $4-20/month for VPS.

Yes, switch models by editing openclaw.json. No restart needed for most changes. Test different models to find what works best for your use case.

Yes, use Ollama integration. Install Ollama, pull models (llama3.3, mistral, codellama), then configure in OpenClaw. Zero API costs but requires powerful hardware (8GB+ VRAM recommended).

Create a .ts file in ~/.openclaw/skills/ with your skill logic. Skills can access the web, run code, control devices, or call APIs. Check our Skills Documentation for templates and examples.

Yes, with custom skills you can integrate with HomeKit, Hue, SmartThings, and more. Use voice commands to control lights, thermostats, locks, and other IoT devices. Requires skill configuration.

Yes, voice calling is supported on WhatsApp and Telegram. Use voicewake to activate with a keyword, or initiate calls directly. Voice responses use text-to-speech. Requires ElevenLabs or similar for high-quality voice.

Yes, OpenClaw has file system access through skills. It can read documents, parse PDFs, generate reports, and write to configured directories. File access can be restricted for security.

Common causes: (1) AI model response time - try faster models like GPT-4o-mini, (2) Slow internet - use closer API region, (3) Low server resources - upgrade RAM/CPU, (4) Too many concurrent requests - enable rate limiting.

Run with --debug flag: openclaw start --debug. Check logs at ~/.openclaw/logs/. Use openclaw doctor to diagnose common issues. Join Discord for community support.

Logs are stored in ~/.openclaw/logs/. Main log is openclaw.log. Use tail -f ~/.openclaw/logs/openclaw.log to watch live logs. Rotate logs weekly with logrotate.

Yes, use custom skills or the HTTP tool. Skills can call any REST/GraphQL API. For webhooks, configure in your channel settings. OpenClaw also supports OpenAI-compatible API for custom integrations.

Yes, configure outgoing webhooks in channel settings. Receive notifications for specific events (new message, voice call, etc.) at your configured URL. Supports JSON payloads and custom headers.

Yes, use webhooks or the OpenAI-compatible API to connect OpenClaw with Zapier, Make, n8n, or any automation platform. Send messages to trigger AI responses or execute actions based on AI output.

Yes, companion apps for iOS and Android connect to your OpenClaw server. Manage settings, view logs, and receive notifications on the go. Download from App Store or Google Play.

Yes, OpenClaw runs on Raspberry Pi 4+ (4GB+ RAM recommended). Use the ARM64 build. Performance is limited, so use lightweight AI models or Ollama with small models like Llama 3.2.

OpenClaw is unique: (1) Self-hosted for privacy, (2) Multi-channel (15+ platforms), (3) Extensible skills system, (4) Full data control, (5) No vendor lock-in. Most alternatives are cloud-only or single-channel.

OpenClaw advantages: (1) Works on WhatsApp/Telegram/Discord you already use, (2) Self-hosted = more privacy, (3) One-time or free AI options, (4) Full customization, (5) No subscription required. ChatGPT Plus is cloud-only with limited integrations.

Get help from: (1) Discord community - fastest response, (2) GitHub Issues - for bug reports, (3) GitHub Discussions - for questions, (4) Our documentation - docs.openclaw.ai. Check existing issues before posting.

Contribute on GitHub: (1) Star and fork the repo, (2) Report bugs or request features, (3) Submit pull requests for fixes or features, (4) Improve documentation, (5) Share your skills in the community. 200+ contributors welcome you!

Yes, enterprise support is available. Contact us for SLA-backed support, custom development, and training. Visit openclaw-ai.com for business plans.