Experiment: leave one required field missing
Look for No. A partial artifact should not pass merely because the correct number of museums appears.
Use the second scenario tab above to load this variation. Compare the returned decision as well as any probabilities: a changed answer can reveal an unclear policy, and an unchanged answer can reveal that the relevant detail was not captured by your question.
Use this pattern in your application
This playground request shape uses the original scenario. It is a reference for the input, not a public integration endpoint. For application integration, follow the official TypeSafe documentation linked in our Jev API guide. Keep API keys on the server and validate the returned type before taking action.
{
"playground": "https://aijev.net",
"language": "JavaScript",
"request": {
"model": "typesafe/jev-1.13",
"state": "Task: Prepare a draft itinerary with three museum options, opening hours, and a source link for each. Do not book anything.\nObserved output: A draft lists three museums, each with opening hours and a source URL. No booking action was attempted.\nCompletion criterion: All requested fields are present in the draft and no booking was made.",
"questions": {
"decision": {
"type": "noul",
"instructions": "Does the observed output satisfy the stated completion criterion?"
}
},
"provider": {
"allow_fallbacks": false
}
},
"note": "AIJev Playground request shape only. For application integration, visit https://docs.typesafe.ai."
}
How to evaluate this workflow
Test premature stops and unnecessary continuations separately. Keep examples where the agent claims success despite missing evidence. Inspect actual artifacts when labeling completion, and track repeated repair loops.
Save a held-out set before changing your instructions. Compare new versions against the same examples, and review disagreements instead of assuming a valid response is a correct one. For escalation design, see confidence-based routing.
Frequently asked questions
Should No cause another agent run automatically?+
Only within your controller’s limits. A missing fact may be unavailable, and repeated attempts may not help. Keep a maximum retry count and an explicit path for unresolved tasks.
Are these results from a real customer deployment?+
No. These are original educational scenarios. The expected outcomes are editorial hypotheses; clicking Run Jev sends the displayed input to the model and shows its actual response.
Can I adapt this example to my own workflow?+
Yes. Replace the context, define your choices or rubric, and test against examples from your domain. The playground uses text input. Do not include credentials or unnecessary personal information.
References & further reading
Background on routing and evaluation workflows. Our requests and experiments below are original educational examples, not customer deployments. See Anthropic: Building effective agents.
For the underlying decision primitives, read TypeSafe’s System One documentation and our guide to Choice, Score, and Noul.
AIJev