Search papers, labs, and topics across Lattice.
VLAA-GUI, a modular GUI automation framework, addresses premature stopping and repetitive loops in autonomous GUI agents by integrating a Completeness Verifier, Loop Breaker, and on-demand Search Agent. The Completeness Verifier enforces UI-observable success criteria, the Loop Breaker filters repeated failures and forces strategy changes, and the Search Agent queries an LLM for unfamiliar workflows. Evaluated on Linux and Windows tasks with five backbones, VLAA-GUI achieves state-of-the-art performance, with some backbones surpassing human performance on OSWorld.
Autonomous GUI agents can now outperform humans on complex tasks, thanks to a novel framework that rigorously verifies completion, breaks failure loops, and searches for solutions.
Autonomous GUI agents face two fundamental challenges: early stopping, where agents prematurely declare success without verifiable evidence, and repetitive loops, where agents cycle through the same failing actions without recovery. We present VLAA-GUI, a modular GUI agentic framework built around three integrated components that guide the system on when to Stop, Recover, and Search. First, a mandatory Completeness Verifier enforces UI-observable success criteria and verification at every finish step -- with an agent-level verifier that cross-examines completion claims with decision rules, rejecting those lacking direct visual evidence. Second, a mandatory Loop Breaker provides multi-tier filtering: switching interaction mode after repeated failures, forcing strategy changes after persistent screen-state recurrence, and binding reflection signals to strategy shifts. Third, an on-demand Search Agent searches online for unfamiliar workflows by directly querying a capable LLM with search ability, returning results as plain text. We additionally integrate a Coding Agent for code-intensive actions and a Grounding Agent for precise action grounding, both invoked on demand when required. We evaluate VLAA-GUI across five top-tier backbones, including Opus 4.5, 4.6 and Gemini 3.1 Pro, on two benchmarks with Linux and Windows tasks, achieving top performance on both (77.5% on OSWorld and 61.0% on WindowsAgentArena). Notably, three of the five backbones surpass human performance (72.4%) on OSWorld in a single pass. Ablation studies show that all three proposed components consistently improve a strong backbone, while a weaker backbone benefits more from these tools when the step budget is sufficient. Further analysis also shows that the Loop Breaker nearly halves wasted steps for loop-prone models.
