Show HN: Tsaagan – Browser control for AI agents that verifies every action (github.com)

Tsaagan, a new project announced by Michael Olmos, aims to empower AI agents with the ability to autonomously navigate the web, filling a critical gap in current AI capabilities. By allowing AI to perform human-like browser tasks—such as logging in, filling forms, and retrieving data from secured portals—Tsaagan streamlines workflows by managing repetitive browser actions. This tool incorporates a robust perception-action-verification process, ensuring that every step taken by the AI is not only executed but also verified for success before proceeding, enhancing reliability and trust in AI automation. This development is significant for the AI/ML community, as it offers a unique, reliable alternative to traditional automation tools, which often lack comprehensive verification methods. Tsaagan utilizes stable grounding through accessibility tree references, allowing it to adapt dynamically to changes in web pages. Additionally, it boasts features like memory storage of learned actions and self-improving capabilities, making it suitable for automation tasks across various domains. Importantly, Tsaagan is designed with ethical considerations, reinforcing its use for authorized tasks while providing transparency in how it accesses and drives web interactions—thus prioritizing user safety and compliance with site terms.