Rob Shocks explains TypeSafe's Jev model, built to output typed decisions rather than generative text. He reviews its core primitives (choice, score, Noul), demonstrates playground examples for triage and smart home actions, and discusses routing use cases.
Original by Rob ShocksGetting startedIntermediate10 min 36 secPublished
The playground examples return a bounded decision and probabilities rather than a generated explanation.
Use Choice for supplied categories, Score for an ordered assessment, and a yes-or-no proposition for a binary probability.
The demonstrated uses include support-ticket priority, smart-home actions, and routing an agent to a tool.
Worth knowing
Speed and cost figures are based on creator demonstrations and vendor claims rather than standardized independent benchmarks; typed outputs provide probabilities without guaranteeing correctness.