Browser Agents Don't Need Better Models. They Need Better Eyes. - Kushan Raj, ARK
Browser agents fail due to poor infrastructure around models, not model intelligence; better page representation makes cheap models fast and reliable.
“models are pretty smart, but it's the infra around them that sucks”
ARK's Kushan Raj argues that browser agents underperform because of weak infrastructure rather than weak models, demonstrating a compressed page representation that lets a cheaper model complete tasks faster than Claude. It matters because it reframes agent reliability as a tooling/environment problem, suggesting cost and latency gains without needing frontier models.