Jev is a decisions model served as an API. Send it some text or data plus a few questions (pick one of these labels, rate this on a scale, is this true?) and it returns a typed answer in the format you asked for, with a calibrated probability for every option. There's no free text to parse and no JSON that breaks.
Use it to route support tickets, score urgency and churn risk, moderate spam, AI slop or abuse, qualify leads, turn scraped text into clean columns, or guardrail AI agents by checking whether a request is in scope or a prompt injection.
Calls cost a fraction of a cent. Pricing is prepaid credits with no subscription, and new accounts get free credits to start.
Question types
- Choice: pick one label from a list you define, such as billing, bug or feature request.
- Score: rate something on a scale you define, such as urgency from 1 to 5.
- Yes/no: answer a true or false question about the input, such as "is this a complaint?"
Every answer comes back with the probability of each option, so you can set a threshold and send low-confidence cases to a person instead of guessing.
How it compares to a general LLM
A chat model writes free text that you then have to parse, validate and retry. Jev returns the exact type you asked for every time, and it is far cheaper and faster per call, so you can run it on every ticket, comment or row instead of a sample. There is a browser playground to try questions without writing code, plus docs and an OpenAPI spec for the REST API.
