# AI use cases for GPT Voice Coding (Simon-AI-coding/GPT-VoiceCoding)

Repo: [Simon-AI-coding/GPT-VoiceCoding](https://github.com/Simon-AI-coding/GPT-VoiceCoding) | Use cases: 3 | Page: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/

A macOS menu-bar app that calls you by voice when a Claude Code or Codex session stops, and sends your spoken or Telegram reply back to that session.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Get a voice call when a coding session stops: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/voice-call-when-coding-session-stops/
- Reply to stopped agent sessions from Telegram: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/reply-to-stopped-sessions-on-telegram/
- Run the bridge engine headless from a config file: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/run-bridge-engine-without-menu-bar-app/

## How to get a voice call when a coding session stops with GPT Voice Coding

Repo: [Simon-AI-coding/GPT-VoiceCoding](https://github.com/Simon-AI-coding/GPT-VoiceCoding) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README

Hear which Claude Code or Codex session stopped and answer by voice, without watching the screen.

**Who it is for:** A developer running several Claude Code or Codex sessions on a Mac who wants to step away from the desk while the agents work.

**The problem:** Agent sessions stop silently, and the developer has to keep checking the screen to find out which one needs input.

**The outcome:** When a session you started yourself stops, the app calls you, tells you what it stopped on, and passes your spoken instruction back to that session.

### Steps

1. Run `curl -fsSL https://raw.githubusercontent.com/Simon-AI-coding/GPT-VoiceCoding/main/scripts/install.sh | sh` in Terminal.
2. Run `codex login` and sign in with your ChatGPT account, since the Live Call uses it instead of an OpenAI API key.
3. Open the app from /Applications and grant microphone access when asked.
4. Start a Claude Code or Codex session yourself in a terminal, because the bridge watches sessions you start and does not launch them.
5. When the session stops, answer the call by voice and say what the session should do next.

### Prompt to paste into your agent

```text
Using the repo Simon-AI-coding/GPT-VoiceCoding, install the macOS menu-bar app on this Mac and start one Claude Code or Codex session in <project-folder>. Report whether the app is running in /Applications and whether a stop in that session triggers a call. Verify by confirming the app appears in the menu bar and that a call is placed after the session stops.
```

### Caveat

The README says the app is macOS only and runs on Apple Silicon, and the Live Call needs Codex signed in with a ChatGPT account.

Risk, remote script: The install command pipes a script from raw.githubusercontent.com straight into sh, so read the script first if you do not trust the source.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/voice-call-when-coding-session-stops/

## How to reply to stopped agent sessions from Telegram with GPT Voice Coding

Repo: [Simon-AI-coding/GPT-VoiceCoding](https://github.com/Simon-AI-coding/GPT-VoiceCoding) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README

Get a stop notice on Telegram while away and reply to it to continue that exact session.

**Who it is for:** A developer who has left the desk while Claude Code or Codex sessions run and needs to unblock them from a phone.

**The problem:** A stopped agent waits idle until the developer returns, even when a one-line answer would let it continue.

**The outcome:** Each stop arrives in Telegram, and a reply to that message reaches the session it came from. Sending /sessions returns the sessions as buttons.

### Steps

1. Run `curl -fsSL https://raw.githubusercontent.com/Simon-AI-coding/GPT-VoiceCoding/main/scripts/install.sh | sh` in Terminal to install or upgrade the app.
2. On first launch, follow the walkthrough to connect Telegram, which is optional and set up there.
3. Start a Claude Code or Codex session yourself so the bridge can watch it.
4. When a stop notice arrives in Telegram, reply directly to that message with your instruction.
5. Send /sessions in Telegram to see the sessions as buttons and pick the one to act on.

### Prompt to paste into your agent

```text
Using the repo Simon-AI-coding/GPT-VoiceCoding, set up Telegram as the second reply route for the installed menu-bar app. Report the Telegram chat state and the list returned by /sessions. Verify by confirming a stop notice appears in Telegram and that a reply to it is delivered to the same session.
```

### Caveat

The README says Telegram is optional and that first launch walks you through it, and it does not describe any other reply channel.

Risk, credentials: Telegram credentials are stored on the Mac by the app, so use a bot and account you control and keep them out of shared files.

Risk, remote script: The install command pipes a script from raw.githubusercontent.com straight into sh, so read the script first if you do not trust the source.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/reply-to-stopped-sessions-on-telegram/

## How to run the bridge engine headless from a config file with GPT Voice Coding

Repo: [Simon-AI-coding/GPT-VoiceCoding](https://github.com/Simon-AI-coding/GPT-VoiceCoding) | Difficulty: Advanced | Time: 1 hour | Basis: From the README

Start the Python engine directly with a config file instead of through the menu-bar app.

**Who it is for:** A developer who wants to run the bridge from a terminal or script on a Mac, for example to test changes to the engine.

**The problem:** The menu-bar app is the default way in, but a developer may need the engine running on its own to inspect or modify it.

**The outcome:** A single Python asyncio engine runs from the command line and watches your sessions without the menu-bar app open.

### Steps

1. Clone the repository with `git clone https://github.com/Simon-AI-coding/GPT-VoiceCoding.git` and change into the folder. (inferred, check the README)
2. Install the package from the repo root with `pip install -e .`, then add the optional audio path with `pip install 'gpt-voicecoding[voice]'` if you need the Live Call. (inferred, check the README)
3. Write a configuration file following docs/control-plane.md in the repo.
4. Run `python -m gpt_voicecoding.engine --config <path-to-config>` in Terminal. (inferred, check the README)
5. Read docs/control-plane.md and docs/adr/0001-hub-and-spoke-bridge-core-with-seams.md to see which seams and adapters the engine uses.

### Prompt to paste into your agent

```text
Using the repo Simon-AI-coding/GPT-VoiceCoding, install the package from source and start the engine with python -m gpt_voicecoding.engine --config <config-file>. Report the engine startup output and the path of the config used. Verify by confirming the process stays running without the menu-bar app open.
```

### Caveat

The README gives the engine command and the voice extra install but not the base install command or the config file format, which live in docs/control-plane.md.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/simon-ai-coding-gpt-voicecoding/run-bridge-engine-without-menu-bar-app/