VLAA-GUI: Knowing When to Stop, Recover, and Search, A Modular Framework for GUI Automation

TL;DR

VLAA-GUI is a modular framework for GUI automation that improves success verification, recovery from failures, and online search capabilities, leading to top performance on benchmark tasks.

Contribution

The paper introduces VLAA-GUI, a novel modular framework with integrated components for stopping, recovering, and searching, enhancing GUI agent robustness and effectiveness.

Findings

Achieved top performance on OSWorld (77.5%) and WindowsAgentArena (61.0%).

Three backbones surpass human performance (72.4%) on OSWorld in a single pass.

All proposed components improve backbone performance and reduce wasted steps.

Abstract

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.