Live Interview Helper is a local-first desktop application. This page describes exactly what data is processed, where it goes, and how to delete it.
Live Interview Helper captures audio from your system's audio output (loopback) and, optionally, your microphone, plus screen content when you trigger Vision Mode. That input is used to transcribe speech, retrieve relevant context, and generate answers inside your own HUD.
Speech recognition and the retrieval engine can run entirely on your machine (local models on Apple Silicon / CUDA / DirectML). Answer generation streams from the AI provider you configure with your own API key — Google Gemini by default, or OpenAI, Groq, or a self-hosted Ollama endpoint. Your API key never leaves your machine: it is stored with OS-level encryption (macOS Keychain / Windows Credential Store) and sent only to the provider you configured.
Transcribed speech and, when you use Vision Mode, screen captures are sent to your chosen AI provider to generate responses. No data is sent to Live Interview Helper servers: the backend runs locally on 127.0.0.1. For fully offline use, configure a local model via Ollama — nothing is transmitted anywhere.
Session transcripts, your API key (encrypted), interview profiles, and settings are stored on your own disk under the application's user-data directory. Log files (backend.log, app.log) rotate automatically and never leave your machine. You can export or delete session history at any time from the Timeline drawer.
The marketing website does not use third-party tracking cookies and does not collect analytics. The download page detects your operating system locally in your browser only to show the correct installer.
Uninstalling the app removes all local configuration. To remove everything immediately, quit the app, then delete the application data folder: %APPDATA%\live-interview-helper on Windows, ~/Library/Application Support/live-interview-helper on macOS, and ~/.config/live-interview-helper on Linux.
Last updated: September 2026 · Questions: open an issue on the GitHub repository.