Every pixel your agent "sees" has a price tag on it.
Before an agent can do anything useful — before it fills a form, clicks a button, or fetches a record — it has to render the page, parse the DOM, and describe every button, banner, and dropdown to itself in tokens. A trivial task burns thousands of tokens before a single productive step happens.
And it's not a one-time cost. Every retry re-reads the same page. Every scroll, every hover, every misfire pays the same toll again.
At demo scale, nobody notices. In production, it shows up on the invoice — and it compounds as you add users, workflows, and retries.
We made the interface cheap for humans and expensive for agents, without ever deciding to. The screen was never meant to be an API.
What task made you realize how much of your agent's token budget was going to UI parsing?
Top comments (0)