Sarathi AI Agent – Open-Source AI Browser Agent Using Structured DOM Reasoning

Summary: Sarathi AI Agent is an open-source browser extension that uses structured DOM snapshots instead of screenshots to perform deterministic multi-step browser actions. It enables contextual Gmail replies, intelligent form filling, and navigation of e-commerce workflows without requiring a backend.

What it does

It injects unique IDs into visible elements to create structured DOM snapshots, allowing a language model to execute actions like click, type, hover, and navigate in a loop until tasks complete.

Who it's for

Developers and users needing AI-driven browser automation with precise control over dynamic websites and multi-field interactions.

Why it matters

It offers a faster, more deterministic alternative to screenshot-based agents, improving debugging and handling complex workflows without backend dependencies.