AdaptArena: Evaluating Test-Time Personalization of Web Agents
It is found that correctly inferring user preferences is necessary but not sufficient for task success, as execution failures in downstream web interactions remain a significant bottleneck even when agents align with the target preference.