FAQ
Common questions about privacy, models, Mac off states, and more.
Is my data private?
Partially. Here's what each part of the system sees:
The TetherClaw relay (relay.tetherclaw.app) routes messages between your phone and your Mac. It can see message content in transit.
Supabase (our database) stores chat history. Every message sent through TetherClaw — both what you send to your agents and what your agents reply with — is logged to Supabase so your conversation history is available in the app. This is what powers the history feature.
Your AI platform (OpenClaw, Hermes, etc.) runs on your Mac. The actual AI model and your agent configuration stay local.
If logging your conversations to a hosted database is not acceptable for your use case, TetherClaw is not the right tool. We're working on options for Pro and Command tier users to self-host the relay and database — those aren't available yet.
What AI models are supported?
TetherClaw doesn't pick or manage AI models. It connects to whatever platform you're running on your Mac. The models available to you are determined by your platform configuration:
- OpenClaw — any model you've configured in OpenClaw (Anthropic Claude, OpenAI GPT, local models, etc.)
- Hermes Agent — supports Anthropic Claude or OpenAI GPT via
~/.hermes/.env - CrewAI / LangChain / AutoGPT — whatever your local setup uses
TetherClaw is model-agnostic. If your platform can handle it, TetherClaw can relay it.
Can I use TetherClaw without OpenClaw?
Yes. OpenClaw is the primary supported platform, but TetherClaw also fully supports Hermes Agent at launch. Additional platforms — CrewAI, AutoGPT, LangChain, and n8n — are coming soon.
The install script handles each supported platform. For Hermes Agent, pass the --hermes flag with your Hermes API key and endpoint.
What happens if my Mac is off?
The app will show Gateway Offline after your Mac has been unreachable for 60 seconds. You can still open the app and browse chat history, but you won't be able to send new messages to your agents.
When your Mac wakes up, the bridge reconnects to the relay automatically. The gateway indicator in the app goes green within a few seconds — no action needed on your part. Any messages you tried to send while the gateway was offline are queued on the relay (up to 100) and delivered as soon as the gateway reconnects.
How is this different from Tailscale?
Tailscale creates a VPN — every device on your Tailscale network can reach every other device directly. That's powerful but requires configuration: installing Tailscale on each device, managing access controls, and typically also building your own mobile UI for your AI agents.
TetherClaw is purpose-built for AI agent access from a phone. There's no VPN, no port forwarding, and nothing to configure beyond the one-line install. The relay handles session continuity, push notifications, approval queues, and the mobile UI is already built. You just connect and start chatting with your agents.
If you need full network access to your Mac from a phone, Tailscale is the right tool. If you want a polished mobile interface for your AI agents with no networking setup, that's what TetherClaw is for.