Overview & result
A local recruiting workspace built on Browser Use's Jev Ultrafast and Browser Harness. Given a hiring brief, Jev chooses where to go on LinkedIn, screens professional titles and assesses visible profile excerpts against the brief, saving profile links with the evidence. TypeSafe's Jev is the default provider and handles navigation, title screening and evidence choices, with no secondary text model.
From the original post
A Jev powered LinkedIn recruiting agent. Watch it browse relevant profiles, save links, and review evidence against your hiring brief.
How Jev fits in the loop
- You give it a hiring brief; it starts from a relevant people search on LinkedIn.
- Jev selects an operation and a compatible target in one request; the executor maps it to an observed element and checks freshness before acting.
- Jev screens visible professional titles before a profile can be opened.
- It assesses visible profile excerpts against the brief and saves the link with the evidence.
How to reproduce
- Inspect the original showcase and source material at https://github.com/skeptrunedev/jev-recruiter
- Verify the bounded prompt and input candidate schema configured for Jev
- Benchmark decision latency and classification accuracy against baseline models
Why this build matters
One provider handles navigation, title screening and evidence choices, and model output never becomes a selector or executable code.
Limitations
- A local workspace: it browses LinkedIn in a live browser on your machine.
- It judges only the profile text that is visible on the page.
- Built on Browser Use's Jev Ultrafast and Browser Harness, so it inherits their requirements.