About Credence

What is this?

This app is not a chatbot — it won't write you a sentence back like ChatGPT or Claude. Instead, you give it one short description of a situation (the state) plus one focused question, and it hands back a simple, calibrated answer: a percentage, a chosen option, or a numeric score — never free-text.

It can only answer three kinds of questions, called primitives:

Yes / No (noul) — a probability from 0 to 1 (e.g. “95% yes”).
Pick one option (choice) — which option best fits, plus the likelihood of each.
Rate on a scale (score) — a number on a scale you define, plus confidence.

The Decision Console page lets you try the same situation and question against three different providers — Jev (TypeSafe AI), Clef (Cloudflare Workers AI), and Perplexity Decisions — so you can compare how they each judge it. Example situations cover both clinical/medical scenarios and non-clinical ones — patient experience, teammate interactions, workplace feedback.

Note: This is a comparison/demo tool, not a medical device. Outputs are model estimates, not clinical advice, and should not be used as the sole basis for a real care decision.