Learn
What is Jev?
A fast decision model: give it state + typed questions, get structured decisions with probabilities — not generated prose.
Traditional LLM
Generate open-ended output
Input → "Write a detailed answer..."
- Writes text and code
- Open-ended output
- Useful for synthesis
- Can be expensive in loops
Jev · decision layer
Choose from a defined answer space
State + Question → Choice / Score / Boolean
- Typed decisions
- Probabilistic output
- No free-form generation
- Designed for high-frequency decisions
Three primitives
The mental model
Most Jev workflows can be expressed with a small answer space.
Choice
Select one option from a developer-defined list.
billing | technical | sales
Score
Rate against a defined rubric or ordered scale.
0 → 1 → 2 → 3
Boolean / Noul
Answer a constrained yes/no decision with probability.
yes 0.97 / no 0.03
Quick example
Support routing
State
"I was charged twice and need a refund today."
Question
Which team should handle this?
billing · technical · sales · other