H-6763: Integrate voice mode into the AI assistant

[kostandinang]

🌟 What is the purpose of this PR?

Unify text and voice inside the existing Petrinaut AI assistant transcript. Voice mode now uses the GPT Realtime migration from the parent PR while preserving the same conversation, Brunch context, and canonical message flow.

🔗 Related links

• H-6763
• Parent migration: H-6763: Migrate voice interviews to GPT-Realtime-2 #9399

🚫 Blocked by

• H-6763: Migrate voice interviews to GPT-Realtime-2 #9399

🔍 What does this change?

• Retires the separate Chat and Interview tabs and replaces the interview presentation contract with provider-neutral Voice mode controls.
• Shows a waveform action when the composer is empty and the normal send action when text is present.
• Renders the active voice session inline in the transcript with live speech, connection/listening/speaking states, a compact divider, and overflow controls.
• Persists voice provenance on ordinary messages and interactive tool output without changing the canonical message path.
• Pauses voice when the panel closes, resumes safely, and supports ordered typed handoff with draft and metadata rollback on failure.
• Adds accessible throttled live announcements, keyboard-operable controls, and reduced-motion fallbacks.
• Updates the user guide, integration documentation, changelog, and changeset for the unified experience.

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

This PR:

• modifies an npm-publishable library and I have added a changeset file(s)

📜 Does this require a change to the docs?

The changes in this PR:

• require changes to docs which are made as part of this PR

🕸️ Does this require a change to the Turbo Graph?

The changes in this PR:

• do not affect the execution graph

⚠️ Known issues

• Voice remains a non-production preview and is disabled by server policy in production.
• Physical-microphone behavior still needs validation in a preview deployment.

🐾 Next steps

• Validate microphone quality, barge-in, panel pause/resume, and typed handoff with a physical device in preview.

🛡 What tests cover this?

• yarn workspace @hashintel/petrinaut test:unit — 246 tests pass.
• yarn workspace @apps/petrinaut-website test:unit — 103 tests pass.
• Focused coverage includes dynamic composer actions, voice provenance, inline state and disclosure, pause/resume, typed handoff recovery, transcript continuity, accessibility announcements, and reduced motion.

❓ How to test this?

1. Configure OPENAI_API_KEY, OPENAI_VOICE_API_KEY, PETRINAUT_OPENAI_VOICE_ENABLED=true, and VITE_BRUNCH_CHAT_ENDPOINT in apps/petrinaut-website/.env.local.
2. Run turbo run dev and open http://localhost:5173.
3. Open AI with an empty composer and select the waveform action; confirm the voice session appears inline in the current transcript.
4. Speak and confirm provisional transcription, listening/speaking feedback, canonical Brunch messages, and voice provenance appear without duplicate messages.
5. Close and reopen the AI panel; confirm voice pauses and can resume.
6. Enter text while Voice mode is active; confirm the composer switches to Send and hands off to text cleanly.

📹 Demo

No static screenshot is attached because the primary behavior is animated and stateful; preview validation should capture a short interaction video.

Made with Cursor

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
hash	Ready	Preview	Aug 31, 2026 11:09pm
petrinaut	Ready	Preview	Aug 31, 2026 11:09pm
petrinaut-docs	Ready	Preview	Aug 31, 2026 11:09pm

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
hashdotdesign-tokens	Ignored	Preview	Aug 31, 2026 11:09pm

[kostandinang]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• H-6763: Integrate voice mode into the AI assistant #9467 👈 (View in Graphite)
• H-6763: Migrate voice interviews to GPT-Realtime-2 #9399
• H-6763: Add the voice interview experience #9380
• H-6763: Improve voice preview reliability diagnostics #9378
• H-6763: Speak finalized Brunch responses with OpenAI #9357
• H-6763: Add OpenAI Realtime voice input to Brunch #9356
• H-6763: Add a generic Petrinaut composer submission API #9355
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 768e1f8. Configure here.}