PulseAugur
EN
LIVE 05:12:04

New HiViG framework boosts AI agents' GUI performance with history and vision

Researchers have developed HiViG, a novel framework designed to improve the performance of Computer Use Agents (CUAs) in complex graphical user interface (GUI) environments. HiViG addresses limitations in existing critics by incorporating both history awareness and visual grounding. The framework's multimodal critic learns from real GUI trajectories to abstract past interactions and evaluate actions with visual context, thereby reducing errors before execution. Testing across web, mobile, and desktop benchmarks showed HiViG significantly outperformed previous critics, improving success rates for models like Qwen3-VL-32B and Gemini-3-Flash. AI

IMPACT Enhances AI agent capabilities in complex GUI interactions, potentially improving automation and user experience across platforms.

RANK_REASON Academic paper detailing a new AI framework and its evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

COVERAGE [1]

  1. arXiv cs.CL TIER_1 English(EN) · Mohit Bansal ·

    A History-Aware Visually Grounded Critic for Computer Use Agents

    Various test-time interventions for Computer Use Agents (CUAs), including critic models, have been developed to improve performance through pre-execution action evaluation in complex Graphical User Interface (GUI) environments. However, existing critics suffer from two key limita…