AutoGUI turns “something looks off” into fixes your agent can apply.
We check both the code and the visuals using visual models. Your agent gets a concrete fix list and finishes the interface by itself.
Web · iOS · AndroidYour agent gets ready fixes, not “something is wrong”
For every problem: where it is, why it happens and how to fix it. The agent applies it and re-checks on its own.
Three steps, none of them yours
The agent builds the UI
Cursor, Claude Code or your own agent assembles the screen.
AutoGUI looks at code and picture
Visual models read both the markup and the screenshot.
The agent gets the fixes
Patches the UI and runs the check again.
Drop a screenshot or HTML and test it for free
Upload an image of a screen or paste the page’s HTML code.
demo form: analysis is being connectedChoose where the analysis runs
The cheaper plan
If you do not need a strict NDA, external visual models do the checking, which keeps the price down.
[PRICE] cheaperNo external models
The analysis runs inside our own infrastructure. Your code and screenshots never reach third parties.
[PRICE] closed perimeterFAQ
How is this better than asking a model to look at a screenshot?
We look at both code and picture: geometry is measured from browser data, and visual models read what code cannot show. Debatable cases are marked “not sure” rather than passed off as fact.
Does my data go to external models?
In Standard mode yes, which is why it costs less. In Private mode the analysis runs without external models, inside our own infrastructure.
Can I use the product today?
Not yet, this is a prototype. We are collecting a waitlist and picking the first teams for a private test.
Show your agent what the result looks like
Leave your email. We will write when access opens and pick a few teams for early testing.