[FEATURE] Add voice conversation mode to Dispatch — talk to Claude, not type to it

Status Closed — not planned
Maintainer reply None cached
Activity 7 comments · opened May 8, 2026 · closed Jul 29, 2026

Feature request

Add a real-time voice conversation mode to the Dispatch UI on iOS (and ideally desktop). Right now Dispatch is a text-message-style interface — type, hit send, wait for text reply. Voice conversation would let users speak to Claude and hear responses back, like a phone call.

Why this matters

For users on iPad-only / accessibility / hands-free workflows, the current "type-and-tap" model is high-friction. A voice mode (similar to ChatGPT's Voice Mode or Gemini Live) would unlock:

  • Long-form planning conversations — verbalize ideas, get pushback, iterate fast
  • Workflow operation while away from keyboard — driving, lying back, walking, recovering from injury (literal use case for me right now)
  • Accessibility — users who can't type comfortably (RSI, mobility, vision)
  • Speed — speaking is 3-4x faster than typing

Proposed UX

  • Toggle in Dispatch header: "Voice Mode"
  • Tap → switches to fullscreen voice UI with waveform indicator
  • Claude speaks responses out loud while continuing to use sub-tasks/tools in the background
  • Visual log of the conversation shown below the waveform for reference
  • Tap to interrupt mid-response (like ChatGPT)
  • Auto-detect speaker silence to commit each turn (no manual "send")
  • Optional speech-to-text fallback for review/editing

Existing reference points

  • ChatGPT Voice Mode — Standard + Advanced (real-time, low-latency)
  • Gemini Live — full duplex, can interrupt
  • Claude Voice Mode (web) — Anthropic already shipped this for claude.ai chat per support article 11101966; just not in iOS Dispatch

Severity / impact

For some users this is the difference between "I can use Claude as my CTO" and "I can't realistically work with Claude today." Voice unlocks AI as a true working partner for non-keyboard contexts.

Related issues

  • #32597 — Voice input not available in Dispatch
  • #36581 — Voice/microphone for Code web interface
  • #30808 — Mic icon in chat input

This is the bigger ask those bugs hint at. Even a tap-to-talk mode (no full duplex) would be a major win.

View original on GitHub ↗

7 Comments

hablutzel1 · 5 months ago

Is there any chance to get something similar to the iOS Siri mechanism (but better)? My intention is to give Claude Dispatch full access to one computer and then while I'm on my AirPods simply say "Hey Claude" and start an interactive conversation with full co trol of my computer

dcpurnell · 5 months ago

Would love to chat with Dispatch on the drive into/from work each day to talk through the day. Having access to Cowork markdown files is the game changer for the voice conversation.

marksoulier · 4 months ago

Looking for a commuting workflow to run introspection with dispatch on the drive, voice mode connection to dispatch would be extremely beneficial. Please add a voice mode to dispatch. This would solve a huge pain point

moesmufti · 4 months ago

I would really like this feature as this would enable it fully as a hermes native solution - I could just talk to it to operate my mac. Right now it feels limited.

mbailey · 3 months ago

Claude Code has been screen-optional for a year now thanks to VoiceMode.

https://voicemode.dev

  • free and open source
  • local voice services
  • works everywhere

AirPods are my primary interface to my laptop

github-actions[bot] · 2 months ago

Closing for now — inactive for too long. Please open a new issue if this is still relevant.

naillizard · 1 month ago

I can't believe this has been closed 😢 .