JevKit

Independent Jev playground & guide

See what Jev decides — and how confident it is.

JevKit is a practical way to understand TypeSafe AI's Jev model: try structured decisions, inspect probabilities, and learn when choice, score, and noul questions fit.

Try the PlaygroundRead the quickstart
One Jev decision
{
  "state": "Checkout fails for EU users.",
  "questions": {
    "priority": {
      "type": "choice",
      "instructions": "What should happen next?",
      "criteria": {
        "fix_now": "Blocker — fix immediately",
        "review": "Investigate before acting"
      }
    }
  }
}

Decision

fix_now

Confidence

0.94

Example shape only. Live Playground results come from the official Jev API.

Built around Jev's actual interface

No fake chat layer. JevKit stays close to the decision primitive Jev exposes.

Typed decisions

Ask a choice, score, or noul question and get a structured answer your software can use.

Probabilities included

Inspect uncertainty directly instead of parsing confidence language out of generated prose.

Editable playground

Change the state, question, and criteria yourself, or start from one of five preset demos.

Safe demo limits

Anonymous users get one live call, signed-in users get five, with basic abuse and cost controls.

Concrete use cases

Explore routing, guardrails, risk scoring, churn signals, and lead prioritization.

Searchable guides

Learn Jev's API shape, question types, and integration patterns without digging through unrelated material.

Jev, in plain terms

The basics behind the playground.

Try one decision end to end.

Load a preset, edit the state, and inspect the exact structured answer returned by Jev.

Open Playground