AI Systems Atlas

G-11Computer-use Coding

/coding-agents/computer-use

The agent drives a real browser to verify UI changes — closing the loop unit tests can’t see.

Step-by-step walkthrough4 steps
  1. The change is to something a test can’t see: layout, color, flow.
  2. The agent drives the actual app the way a user would.
  3. The screenshot goes back into the model — vision closes the loop.
  4. “Looks right” becomes a verifiable step, not a hope.
Key insight

A screenshot in the loop makes “looks right” a checkable claim: vision extends the verifier to layout, styling, and flow.

Failure mode

Verifying single frames only: the bug that needs a hover, a scroll, or a second click stays invisible.

Further reading
RelatedTest loopCI feedback loop

Elsewhere in the atlas

Harnesses →Security →Evals →Context →Tools →Ops →