Personal Graphical User Interface (GUI) agents depend on a user's interaction history to interpret ambiguous instructions and anticipate recurring routines. A common approach retrieves task-relevant history and appends it to the policy's context, assuming that experience relevant to a task stays useful for every decision within it.
A new paper challenges that assumption. The authors find that retrieved history strongly improves the opening step of an episode but delivers little sustained benefit across the remaining 90% of steps, and offers weak guidance on whether a proactive suggestion is warranted. A relevant record may indicate where to begin, but not which past action fits the current screen or whether a routine is due now.
The paper frames the problem as a distinction between relevance and applicability: relevance is determined at the task level, whereas applicability depends on the situation at decision time. The authors therefore recast personalization as experience activation and introduce ExpActivator, a training-free framework that activates only the experience applicable to the current situation.
During execution, ExpActivator matches each new screen to historical states in the latent space of a frozen GUI backbone and supplies the corresponding action as a reference. Before execution, it activates a recurring intent only when the current time and scenario provide sufficient support, and otherwise abstains.
Across four GUI backbones, the framework improves within-trajectory step success by 28% on average, achieves the best personalized execution on every backbone while using roughly one-fifth as many history tokens, and reaches about 2.3 times the Matthews correlation coefficient of the strongest proactive baseline. As the authors put it, experience pays where it is activated, not where it is appended.