Year: 2026
Type: conference-paper
Abstract: End-to-end GUI agents for real desktop environments require large amounts of highquality interaction data, yet collecting human demonstrations is expensive and existing synthetic pipelines often suffer from limited task diversity or noisy, goal-drifting trajectories.We present a trajectory expansion framework ANCHOR that bootstraps scalable desktop supervision from a small set of verified seed demonstrations.Starting from each seed, we identify branch points that correspond to meaningful state changes and propose new, stategrounded task variants conditioned on the current GUI context.An executing agent then follows the proposed instructions to generate new trajectories, while a verifier enforces task completion via state-aware checks and trajectorylevel consistency.To improve supervision quality, we further apply task-conditioned steplevel filtering to remove ungrounded actions and denoise post-branch segments to maintain coherent intent.Experiments on standard desktop benchmarks, OSWorld and Win-dowsAgentArena, show that models fine-tuned on our expanded corpus achieve consistent improvements over zero-shot agents and representative synthesis baselines, and generalize across applications and operating systems.
Authors: Jinbiao Wei, Yilun Zhao, Kangqi Ni, Arman Cohan
Institutions: University of North Carolina at Chapel Hill, Yale University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7166850759