Install

Interpreter is a desktop app. Download it, point it at a folder, pick a model, and it can start driving your real apps.

Download

Grab the build for your OS from the download page.

• macOS (Apple Silicon and Intel)
• Windows
• Linux

Open the installer and launch the app. There is no account to create.

First launch

The app walks you through three things in order:

• Pick a workspace folder. This is the only directory Interpreter can read or write in by default. Use a real working folder, not your home directory. You can change it later or add more folders from settings.
• Pick a profile. A profile is a provider plus a model plus credentials. Choose one of the built-in providers (OpenAI, OpenRouter, Ollama, LM Studio, NVIDIA, or any OpenAI-compatible endpoint) and paste in a key or base URL. See Connect a modelConnect a modelPreview this page..
• Grant OS permissions. Interpreter controls real desktop apps, so the OS will prompt for it.

OS permissions

Be honest with the prompts. The agent cannot click, type, or read on-screen state without them.

• macOS: grant Accessibility and Screen Recording in System Settings. Restart the app after granting.
• Windows: approve the UAC prompt on first launch. Some apps require running Interpreter as the same user that owns them.
• Linux: depends on your desktop environment; the app will print what it needs.

If you skip a permission, the overlay will tell you which one is missing the first time it tries to use it.

Voice mode

Voice mode auto-installs a local speech-to-text model the first time you launch the app. You do not need to install anything by hand. It runs offline once installed.

What to try first

Open the app, point it at a folder with a few real files, and run through Your first taskYour first taskPreview this page.. Pick something boring you already do by hand.