Why 99% Accurate Browser Agents Still Fail — Derek Meegan, Browserbase
Browser agents that work in demos routinely fail at production scale due to compounding error rates
“production is a completely different matter. Thousands of unattended runs while the site may change, or your model may just be having a bad day.”
Derek Meegan of Browserbase explains the demo-to-production gap for browser agents, where high single-step accuracy still compounds into frequent failures across thousands of unattended runs. He outlines three browser interaction strategies—text representation, computer use via screenshots, and agent offloading via dynamic code execution—and why most production systems rely on the first two. The talk highlights a critical infrastructure challenge as enterprises move browser agents from proof-of-concept to reliable automation.