Jul 2026
StateAct: Program State, before Pixels, for Long-Horizon Computer-Use Agents
Grounding action, verification, and memory in state, what the authors call state-grounding, shifts the main bottleneck from perception toward reasoning: failures depend more on what the agent thinks than on what it sees.
Yan Yang, Xiang-Ru Jian, Ziyang Luo et al.
· arXiv.org · 4 citations