Interact with AI agents using voice commands or text. This client offers on-device speech recognition for privacy and streams responses with markdown rendering. It supports multi-agent channels, image sending, and direct interaction with agent tools.
1.1.0
May 18, 2026
Conversation Mode is here. Tap once, talk, and ClawTalk handles the rest — on-device transcription, streaming reply, and a spoken response. Then it keeps listening for your next turn. Hands-free, end-to-end. Push-to-talk and text input still work the same.
MoreNo, ClawTalk does not contain advertisements. It is a paid application available for a one-time purchase price of $2.99.
ClawTalk is currently supported on iPhone devices. It is designed to leverage the Apple Neural Engine for on-device speech recognition.
The latest version of ClawTalk is 1.0, released on March 23, 2026. As an open-source project, updates are driven by community contributions and development efforts.
ClawTalk has an age rating of 4+. This indicates it is suitable for a wide range of users, including children, with no mature content concerns.
While the app itself is available offline, its functionality relies on connecting to a self-hosted OpenClaw gateway server. The Apple TTS provider can work offline for spoken responses.
Chat history in ClawTalk is encrypted at rest using iOS Data Protection. All credentials are also securely stored in the iOS Keychain, protected by the Secure Enclave.
To use ClawTalk, you need a running OpenClaw gateway server with the HTTP API enabled, a secure connection (e.g., Cloudflare Tunnel or Tailscale), and a gateway token for authentication.
Yes, ClawTalk allows you to attach up to 8 photos per message for visual context. Your AI agent can then see and respond to these images.