How can I help you today?
Log in to save your chat history across visits.
Small On/Off switches for camera & mic
Don't have an account?
Upload PDF, TXT, MD or DOCX files. Ferrari AI chunks and indexes them so chat can answer questions from them.
Upload an image for real analysis — dimensions, format and dominant colors via GD. No object/scene recognition (that needs an AI vision model this project doesn't have).
Say "remember that ..." in chat, or add a fact here directly. Ask "what do you remember about me" to recall one, or "forget ..." to remove it.
Toggle a tool off to stop Ferrari AI from using it. Calculator and dates work directly in chat — try "12 * (5+3)".
Give a goal with steps joined by "then" — e.g. "calculate 12*8 then search the web for the eiffel tower". Each step runs a real tool; a step that matches nothing is skipped, not guessed.
Same goal syntax as the Agent Engine, run automatically on a schedule. Automations run while this dashboard is open (it checks periodically) — see backend/README.md for wiring real background execution via Windows Task Scheduler.
Create an OpenAI-compatible bearer key for Cursor, apps and scripts. Keys provide access to your Ferrari AI quota—never publish them or put them in browser-side code.
Base URLLoading…
Recommended modelferrari-ai
Ferrari AI never runs shell commands on your machine. A separate local daemon (backend/desktop_agent.php) polls for only the specific, permission-gated actions you allow below.
Ferrari AI’s Eye can only look after you allow camera access. Permission is requested every time before seeing — nothing is recorded until you capture.
Preview below. Capture when ready. Camera turns off after capture or when you close.