Intern: an AI agent that uses this desktop in plain sight

Intern is an AI agent that uses this site’s desktop in plain sight: given a task, it opens apps, clicks, picks options, types and reads with a visible cursor, one action a step, chosen by Llama 3.3 70B on Workers AI from a text snapshot of the desktop’s controls. It can’t send messages or open other sites, asks before any download, logs every step with what it saw and why, and checks its answers against what was on screen.

How it works

  1. Pick a task card or type your own, and press Start.
  2. Each step it reads a text snapshot of the desktop (the open windows, the front window’s text and its controls) and does one thing with its own cursor. Every step joins a timeline: what it did, where, and what it was thinking.
  3. Its answer comes with checks and sources: a number that never appeared on screen is flagged, and each source brings its window forward.

Privacy and limits

  • You’re in control: pause it, press Esc or click outside its window to take over, and resume from where it stopped.
  • It can’t send messages or voice memos, empty the Trash, or open other sites or new tabs, and it asks before downloading a model to your device.
  • Each step, your task and the text snapshot go to Llama 3.3 70B on Workers AI.
  • It can misread the screen or stop short: check its answer against the window it used.

Intern runs on this site’s desktop, which needs JavaScript: with it on, this address opens the app. In the desktop’s terminal it’s app intern. This page as Markdown.

Finder