Use Codex by voice from your own app shell

They wrapped Codex in custom phone and desktop clients so they could speak requests, hear answers, and keep the same flow on both devices.

Done with Codex· Advanced

What they did
They built a coordinated desktop app and phone app around the Codex App Server, with both clients pointing at one backend so the experience stayed consistent. Spoken input was transcribed first, then forwarded to Codex, and the reply came back as audio through OmniVoice or browser speech. On top of that, they added nicer answer display, document attachment, auto-resume, and a handoff to another business seat when one seat ran out of capacity.
How it went
They said the setup was a big improvement over the CLI and made it easy to switch from phone to desktop without relearning anything, while Codex kept talking in the background. They also noted that this was expensive and work-intensive; it only makes sense when you need substantial customization, and they were already on 10 business seats.
Tips
Treat this as a custom integration project, not a quick tweak. If you try it, build the basic speech-to-text and spoken reply loop first, then add the extra features later.

Try it yourself with Codex

Help me build a [desktop or mobile] app that uses Codex through the Codex App Server for [my use case]. I want to speak into the app, have my speech turned into text, send that text to Codex, and hear the response read back. Add [attachments / auto-resume / better response display / seat fallback] if helpful. Before any action that would send, submit, delete, or charge money, stop and ask for my approval. Start by outlining the screens, data flow, and the smallest version I can test first.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on reddit.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.