SpeakAble
Voice-first AI assistant for iOS. Talk naturally to your Hermes agent — hands-free, private, and self-hosted. Like Grok, but yours.
Coming to TestFlight
Voice-first AI assistant for iOS. Talk naturally to your Hermes agent — hands-free, private, and self-hosted. Like Grok, but yours.
Continuous conversation mode — just start talking. SpeakAble listens, your Hermes agent thinks, and you hear the response out loud. Like a phone call with your AI.
Interrupt anytime. If SpeakAble is speaking and you want to jump in, just tap the mic — it stops instantly and listens. Natural, like real conversation.
Runs against your own Hermes or OpenAI-compatible API. LiveKit server is free (Apache 2.0). No subscriptions, no cloud lock-in. Your agent, your data.
Speak in English, Cantonese, or Mandarin. Apple on-device speech recognition and synthesis — all voice processing stays on your iPhone.
Every conversation is saved with full text transcripts. Switch between sessions, review past discussions, and pick up where you left off.
Hermes can respond with tables, code blocks, headings, and bullet lists — all rendered beautifully in dark mode on your iPhone.
You provide the brain (Hermes or any OpenAI-compatible agent). SpeakAble handles voice I/O on your iPhone. All connected through a free LiveKit server.
One command: brew install livekit. Free, open-source, runs anywhere.
Thin Python proxy connects LiveKit to your Hermes API. 4 dependencies, zero config.
Enter your server address once in Settings. SpeakAble auto-derives LiveKit and token endpoints.
Tap the mic, speak naturally. Your Hermes agent responds out loud. No typing, no screens.