USE WHEN
When to use this experiment
You want to compare a claim with evidence you already have.
CLAIM / EVIDENCE · R06
Enter a claim and evidence you provide to see whether the evidence supports, contradicts, or does not address it. The tool does not search the web.
YOUR INPUT
Edit the example input
RESULT
Jev’s answers
Provider confidence: 55%
The model evaluates. Deterministic policy code decides the displayed application path.
QUESTIONS FOR JEV
Edit the questions
Editing a question requires a new model run. Each experiment keeps its answer types, options, and scoring scale fixed.
APP RULES
Change how your app uses the answer
↯ Changing these rules updates the result in your browser without another model call.
{
"origin": "fixture",
"answers": {
"relation_e1": {
"type": "choice",
"choice": "supports",
"probabilities": {
"supports": 0.76,
"contradicts": 0.07,
"irrelevant": 0.05,
"insufficient": 0.12
}
}
},
"providerConfidence": {
"relation_e1": 0.55
},
"policyDecision": {
"action": "supported",
"label": "Supported by supplied text",
"reasonCodes": [
"support-above-threshold"
]
},
"run": null
}import { experimental_evaluate as evaluate } from 'ai';
const state = {
"claim": "The customer was refunded.",
"evidence": [
{
"id": "e1",
"text": "Synthetic support transcript: We returned the full payment to your original payment method today."
}
]
};
const questions = {
"relation_e1": {
"type": "choice",
"instructions": "What is the relationship between supplied evidence e1 and the claim? Judge only the supplied texts, not outside-world truth.",
"criteria": {
"supports": "The supplied text directly supports the claim.",
"contradicts": "The supplied text directly contradicts the claim.",
"irrelevant": "The supplied text is unrelated to the claim.",
"insufficient": "The relationship or support is ambiguous or incomplete."
}
}
};
const policy = {
"minRelationProbability": 0.7
};
const result = await evaluate({
model: 'typesafe-ai/jev',
state,
questions,
maxRetries: 0,
providerOptions: { gateway: { zeroDataRetention: true } },
});
// Validate result.answers before applying your deterministic policy.
console.log({ answers: result.answers, policy });
USE WHEN
You want to compare a claim with evidence you already have.
LIMITS
You need fact-checking, source discovery, or a legal/scientific truth verdict.
BUILT-IN CHECKS