Tag: #web-automation7 results
The researchers introduced State-Grounded Dynamic Retrieval (SGDR), a method that enables web agents to retrieve and reuse skills dynamically based on both task goals and current webpage states. This approach addresses the limitations of static skill retrieval, resulting in significant performance gains on the WebArena benchmark.
Opera’s latest update introduces the MCP Connector, transforming Neon into an autonomous browsing agent. This allows AI models like ChatGPT and Claude to directly control the browser. Users can now automate tasks across websites seamlessly.
Hark previewed a browser-use agent designed to complete tasks on the web. The company claims its agent is faster and cheaper than competing offerings.
BrowserBC introduces a mechanism to record human web interactions once and replicate them across any AI agent. This simplifies the automation of complex web workflows.
Google has discontinued Project Mariner, an experimental AI feature for autonomous web tasks. Shut down on May 4th, 2026, its technology is now integrated into other Google products like Gemini Agent. First revealed in December 2024, it was updated to handle up to 10 tasks simultaneously.
Tests Google's Auto Browse, turning Chrome into an AI-agentic browser. Evaluates performance on shopping, research, and emailing tasks.
Chrome's Auto Browse agent autonomously surfs the web with impressive results. However, it also experiences spectacular crashes. Testing reveals both strengths and reliability issues.