Data & research built with Jev

Grok 4.7 proposes the moves, Jev makes the fast state decisions in under 0.3 s, and 20+ futures of a cancer case play out in minutes.

1,500+ comments become 10,000+ small Jev decisions, evaluated in parallel to surface what the audience actually thinks.

Jev classifies thousands of listings by architecture, renovation status or distance to freeways in under 20 seconds, for $0.18.

Proq's classifier matched GPT-4.1 on every sheet for $0.0052, 17–21x cheaper and 5x faster than their production pipeline.

OCR plus Jev: a little classifier that sorts a folder of images into categories.

Jev reads every book index in parallel in 300 ms: "Who invented photography?" points to page 98, under Niépce.

Paper Breakdown adds Jev on top of its recommender: ~70 picks per user for $0.0019, about $5 a month.

Jev classified 100 emails in 1.42 s and sent the uncertain ones to Kimi K3: 96/100 correct for ~$0.07. Not sped up.

Stockfish analyzes the moves, Jev finds recurring patterns across your Chess.com games, and a dashboard shows what correlates with losses.

Jev reads every comment on the most-discussed threads for stance, substance and open questions, then a verdict card sums it up.

Jev on the Venice API sorts 24,000 HN posts into 12 categories in about two minutes.

No Sugar For Kids scores every snack on several criteria with Jev, for $0.11.

Helps systematic reviewers find and extract information from research articles quickly. Open source.

A pathfinding showdown: can Jev's step-by-step choices keep up with classic shortest-path algorithms?

Jev QA'd egocentric action labels for physical-AI training data, for 90 cents.

During a consult, Jev reads the ambient scribe's transcript and updates symptoms, the differential diagnosis and red flags in real time.