OpenClaw FAQ

OpenClaw is an open-source AI assistant that runs on your own server. It connects to messaging platforms (WhatsApp, Discord, Telegram, Slack, iMessage), automates tasks, and gives you a personal AI that works 24/7. Unlike cloud AI services, you control your data and costs.

About 15 minutes if you use a VPS with Docker pre-installed (like Hostinger's 1-click deploy). If you're setting up from scratch on a fresh server, budget 30-45 minutes for the full setup including Docker installation.

No. The web dashboard lets you configure integrations, set up automations, and manage your AI assistant without writing code. Advanced users can extend functionality with custom scripts, but it's not required for basic use.

ChatGPT is a hosted service — your data goes through OpenAI's servers. OpenClaw runs on YOUR server, connects to YOUR apps (WhatsApp, Discord, etc.), and can use ANY LLM (OpenAI, Anthropic, local models). You control the data, the integrations, and the costs.

Yes, but it won't be available when your laptop is asleep or offline. For 24/7 access and reliable integrations (especially WhatsApp which needs constant connectivity), a $6-7/month VPS is strongly recommended.

OpenClaw itself is free and open-source. Your costs are: (1) VPS hosting: $6-7/month for a basic server, (2) LLM API: $5-20/month depending on usage. Most users spend $10-25/month total for a personal AI assistant running 24/7.

For light usage, OpenAI GPT-4o-mini offers the best value at ~$0.15 per million input tokens. For heavy usage, Groq offers free tier access to Llama models. For privacy, local models (Ollama) are free but require more RAM/CPU.

The main variable cost is LLM API usage. If you connect WhatsApp, you might want a dedicated phone number (~$2-5/month from Twilio). Some integrations (like Notion API) have their own rate limits on free tiers. Otherwise, it's predictable: VPS + API credits.

Almost. OpenClaw is free software. You can run it on an old laptop (free), use Groq's free tier or local models (free), and use free integrations like Discord. The trade-off: no 24/7 uptime, slower responses, and more technical setup.

Yes. OpenClaw runs on your server — your messages, files, and data never leave your infrastructure. If you use OpenAI/Anthropic APIs, only the prompts you send go to them. If you use local models (Ollama, LM Studio), nothing leaves your server ever.

Essential steps: (1) Change the default GATEWAY_TOKEN to a long random string, (2) Set up a firewall (ufw allow 22, 3000), (3) Use HTTPS with Caddy or Nginx, (4) Keep Docker updated. Full walkthrough in our security checklist.

Only if you allow them. By default, OpenClaw binds to localhost. To expose it publicly, you'd need to configure port forwarding and ideally add authentication. The gateway token acts as a basic API key.

Messages are processed by OpenClaw on your server and stored locally in your data volume. They're not sent to third parties unless you explicitly configure an integration that does so. You can delete the data volume to wipe all history.

WhatsApp, Discord, Telegram, Slack, iMessage (via BlueBubbles), Instagram DM, Facebook Messenger, and more. Over 50 integrations are available. Each has its own setup guide.

OpenClaw uses the WhatsApp Web protocol. In the dashboard, go to Integrations > WhatsApp and scan the QR code with your phone (Settings > Linked Devices). Your phone needs internet for the connection to work, but doesn't need to stay online 24/7.

Yes. You can have OpenClaw respond on WhatsApp, Discord, and Telegram simultaneously. Each integration has its own configuration, and you can set different behaviors per channel if needed.

Yes! The Home Assistant integration lets OpenClaw control smart home devices, read sensors, and trigger automations. You can ask 'what's the temperature' or 'turn off the living room lights' through any connected messaging platform.

Yes. Both integrations allow OpenClaw to search, read, and (optionally) modify your notes. This lets you ask questions about your knowledge base or have the AI help organize your notes.

Usually a configuration error. Run `docker compose logs openclaw --tail 50` to see the crash reason. Common causes: missing API keys, invalid docker-compose.yml syntax, or out of memory errors.

First, check if the container is running (`docker compose ps`). Then test locally (`curl http://localhost:3000`). If local works but external doesn't, it's a firewall issue — open port 3000 in your cloud provider's firewall settings.

You have about 30 seconds to scan. Open WhatsApp on your phone first (Settings > Linked Devices > Link a Device), then refresh the QR in OpenClaw and scan immediately. Also check you're not at the 4-device limit.

Check for typos or extra spaces in your .env file. The key should start with `sk-` for OpenAI. Also verify the key has the right permissions at platform.openai.com/api-keys. After fixing, restart with `docker compose down && docker compose up -d`.

Yes. Install Ollama on your server and configure OpenClaw to use it. Local models like Llama 3.1 or Mistral work well for general tasks. You'll need at least 8GB RAM for decent performance with 7B parameter models.

Run `docker compose pull` to fetch the latest image, then `docker compose up -d` to restart with the new version. Your data persists in the mounted volumes. Check the release notes for any breaking changes before updating.

Yes. Each instance needs its own data directory and port. Use separate docker-compose.yml files or define multiple services in one file. This is useful for separating personal and work assistants.

Yes. OpenClaw has a skills system that lets you define custom actions, prompts, and behaviors. Skills can be written in TypeScript/JavaScript and loaded from the data/skills directory. Advanced users can create sophisticated automation workflows.