Jev targets fast, low-cost structured decisions

Jev, TypeSafe AI’s new decision model, returns type-safe structured values such as a choice, a score, or a probability instead of generated text. The model is priced at 4 cents per million input tokens with no output charge, making it useful for classification-heavy workflows where conventional LLM calls can be too expensive or slow.

Claire Vo tested Jev across five real projects: PR categorization, analysis of Claude and Codex sessions, Gmail triage, the ChatPRD product insights graph, and a live dashboard built from 4,500 YouTube comments. One workflow analyzed 1,700 PRs for 9 cents to better understand how engineering effort was distributed.

The projects highlight Jev’s role as a classification and scoring layer rather than a standalone assistant. In Gmail triage, Jev scoring was paired with LLM follow-up, while the ChatPRD product insights workflow used 1,100 signals and 200,000 classifications. Additional demos included a searchable audience signal dashboard and a real-time voice-to-color emotion-mapping app.