In plain language
How it works
AI Fit watches what you ask for — a source, a comparison, a quick answer, or a hands-off helper — then matches that to a dated list of products and models. A chat model may label extra notes. It never picks the winner.
1. Short scenes, not a personality test
Each choice becomes a simple signal. Those signals turn into a 0–1 picture of how you like to use AI, with confidence from how often we saw it. The ranking is math, not a vibe.
2. Products and models stay separate
Apps and models are different catalogs. Every public row has a last-checked date. Stale rows lose confidence; they are not marked “bad.”
3. You can see why
Results quote the choices that drove each dimension. AI Fit is similarity, not a scientific probability. You can save or delete the session from your results page.
4. Version
AI Fit 0.4. Usually four scenes (about five minutes), continuing only when a core habit still is not clear. Scoring stays on the server. The catalog last-reviewed date is derived from each row's last_evaluated_at and shown on the product and model registry.
5. Limits
Scores are not scientifically validated. This is not a personality test, clinical tool, or hiring screen. Do not use it that way.
AI Fit is a personal fit quiz. It is a separate product from AI Fit Teams (a workload assessment) and from ARK Control (production measurement). This page does not send you there.