What can you build with Jev? | Rubric Labs

Written by AI

TLDR: Jev returns probabilities over predefined answers in milliseconds, enabling notification triage, semantic search, spellchecking, settings navigation, and iterative email rewriting.
Best for: Fast, repeated decisions over predefined options
The tip: Write explicit criteria, provide the relevant context, and let code enforce the confidence threshold.
What you need: Predefined possible answers
Relevant state or context
A confidence threshold in code
Start here: Choose Choice, Score, or Bool for the decision
Write the criteria in plain language
Test thresholds against labeled examples
Filed note: Rubric Labs tests Jev, a fast decision model that scores predefined answers, across notification triage, semantic search, spelling, settings search, and email rewriting.

Highlights: More precisely, it acts as a classifier or scorer: it returns probabilities for predefined answers, and your code determines what action to take.
The model scores options while application code controls the action.
One question rarely covers everything you need to know, so it often pays to ask several in one request
Pair questions when one score cannot establish both relevance and certainty.
What’s different about Jev is that it’s incredibly fast, cheap, and consistent
Its speed and cost open up per-item decisions that larger models make impractical.
Excerpt: Jev is a new model built to evaluate questions with predefined possible answers. It returns answers in a few hundred milliseconds, in a structured format, with a probability for each possible answer.
From rubriclabs.com