Your practice. Your choice.
A quieter app should be clear about what happens to your words.
How your coach works
- This practice uses
- Hosted AI
- Server configuration
- hosted provider · configuration present
Your reflection and relevant session context are sent by this application’s server to the configured hosted model provider. Provider retention policies apply. Do not include sensitive personal information.
The provider is selected by the server operator through environment variables, not by a browser toggle. API keys are never sent to your browser. If a model fails, we say so.
Your observations can stay on your device.
When you run the app, its database, and an open-weight model locally, your observations stay on that machine. In this hosted preview, account information and saved observations are stored in the configured MongoDB database (normally Atlas for hosted deployments)—not exclusively on your phone.
No analytics. No tracking. No public profiles.
Each account can access only its own observations and skill history.
Optional browser voice recognition may send audio to the browser vendor. We ask before enabling it. Listening stays open until you stop it or after about 5 seconds of silence. You can always type instead.
For developers · Configure local or hosted coaching
Run Next.js with a MongoDB replica set on your machine for fully local use — either your already-installed MongoDB with its replica set enabled (no Docker; see DATABASE.md or run scripts/mongodb-enable-replicaset) or the supplied Docker Compose service. Set these values in .env.local and restart the server. The URL is reached by the application server, not the browser.
# Local open-weight model LLM_PROVIDER=local LOCAL_LLM_BASE_URL=http://localhost:1234/v1 LOCAL_LLM_MODEL=your-model # Or an OpenAI-compatible hosted provider LLM_PROVIDER=hosted HOSTED_LLM_BASE_URL=https://your-provider.example/v1 HOSTED_LLM_MODEL=your-model HOSTED_LLM_API_KEY=server-side-secret
Choose LM Studio, Ollama, llama.cpp, or another compatible endpoint. Check README.md and architecture.md in the project for setup details.