GoKawiilTypeSafe releases Jev Ultrafast, an open-source browser agent using indexed action targeting
TypeSafe has open-sourced Jev Ultrafast, a browser automation agent that reads a page's elements as an indexed table and picks an operation-target pair in a single network round trip, calling a small language model only when text needs to be generated. In a demo, it booked a Zürich-to-London flight search on Google Flights from a single natural-language goal in 7.1 seconds, including page load waits. The code, along with a local inspector interface for watching element probabilities and executed actions, is available on GitHub.
GoKawiil's interpretation of the reporting above, not reported fact.
Most browser agents rely on large models reasoning over screenshots or DOM dumps, which is slow and expensive; Jev's approach of constraining actions to a dynamically generated, verified set of valid targets could make agentic web automation far cheaper and faster to run at scale. Because there are no site-specific scripts, the same architecture could generalize across many web services without custom engineering per site, a key barrier for practical AI agents.
- Jev separates action selection (fast, indexed) from text generation (small LLM), cutting latency dramatically.
- A single flight search demo completed in 7.1 seconds with no hardcoded site logic.
- The project is open source and runnable locally via a Chrome-connected inspector interface.
Source: github.com, 2026-09-17
Published there as: “Jev Ultrafast: A browser agent with a dynamic, indexed action space”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.