How it works
Rexe turns "trust me, I can do this" into something an employer can actually rely on — without pretending to be more certain than it is.
1. An adaptive interview
You answer a short, conversational interview in your browser. It's not a quiz with right answers — it asks about real work you've done and follows up on your own answers, digging for concrete specifics: the project, what broke, what the logs said, what you tried. A real practitioner gets more fluent as the questions get more specific; someone at their ceiling gets vaguer. The interview reads that and moves on — it won't grind you on something you've said you don't do.
2. It picks the right track
Early on, it works out whether you're a deep systems engineer, an AI-assisted builder, or a blend — and probes the competencies that actually matter for you. A builder isn't marked down for not knowing low-level internals; an engineer isn't marked down for not having shipped a product solo.
3. Honest grading
A separate, more thorough model reviews the whole transcript and produces a calibrated report. Every competency gets a status:
- Verified — backed by concrete specifics that held up under questioning.
- Partial — correct, but thin on lived experience.
- Claimed only — asserted, with no depth or evidence behind it.
- Unconfirmed — not covered.
There's no single "pass". The report names your strengths and your gaps — because showing the gaps is exactly what makes the strengths believable.
4. A report you can share
You get a clean report with a shareable link. Hand it to an employer or recruiter; it's designed to be a credible signal precisely because it doesn't overstate.
How honest is it about being gamed?
Very. No async text interview is perfectly cheat-proof, so Rexe doesn't pretend otherwise. It defends against copy-pasting answers from an AI by demanding lived specifics tied to your own work and following up unpredictably, and it calibrates confidence accordingly. Higher-assurance modes (voice, and proctoring for high-stakes use) are on the roadmap — the report always tells you the assurance level rather than implying a guarantee.