Free tier
Send us your text, get a decision back
A key, one endpoint and eight rulesets that were tuned and measured before you arrived. You do not write question shapes, pick a model, or learn that a score is a weighted mean of rubric indices rather than a 0-1 float. That translation is the thing.
Quick answer · verified 2026-09-19
What is this, and how is it different from calling Jev directly?
Calling Jev directly means writing Choice, Score and Noul objects, discovering that option descriptions are the entire prompt surface, and finding out the hard way that score runs 0..n-1 rather than 0..1. That work is already done here.
You post text and a ruleset name. The rulesets are generated from the fixtures in our public benchmark, so the cost and latency published on this site are the numbers these exact questions produced — not a marketing approximation of them.
| Free tier | 1,000 decisions / month — one decision = one question about one item |
|---|---|
| Signup | an email address — no password, no account, no confirmation mail |
| Rulesets | 8 — the same ones measured in our public benchmark |
| You send | your own text — not Jev question objects |
Make a call
curl -X POST https://jev-agent.com/api/v1/classify \
-H "Authorization: Bearer jv_live_…" \
-H "content-type: application/json" \
-d '{
"ruleset": "support-ticket-triage",
"items": ["Charged twice for September and cancelling Friday unless refunded."]
}'{
"ruleset": "support-ticket-triage",
"ruleset_version": 1,
"results": [
{ "index": 0,
"answers": {
"queue": { "choice": "billing", "confidence": 1, "probabilities": {…} },
"severity": { "score": 2.5, "legend": {"0":"Low","1":"Normal","2":"High","3":"Urgent"} },
"churn_risk": { "noul": 0.95 }
},
"latency_ms": 1669 }
],
"usage": { "decisions_charged": 3, "used": 3, "remaining": 997 }
}The rulesets
Up to 20 items per request. Billing counts decisions, not requests: a ruleset asking three questions spends three per item, because that is what the upstream call actually costs. The catalogue is also readable without a key at GET /api/v1/classify.
| Ruleset | What it decides | Per item |
|---|---|---|
| support-ticket-triage | Route an inbound ticket to a queue, grade its urgency, and flag churn risk. | 3 |
| llm-model-routing | Decide how much model capability a request actually needs, before you spend it. | 1 |
| agent-tool-selection | Pick which tool an agent should call next from a fixed toolset. | 1 |
| content-moderation | Grade harm on a four-band rubric and flag targeting and spam separately. | 3 |
| phishing-detection | Independent yes/no signals for an email — combine them yourself rather than trusting one verdict. | 5 |
| rag-reranking | Score a retrieved passage for real relevance, not embedding similarity. | 1 |
| agent-output-guardrails | Verification checks on a draft response before it reaches a user. | 4 |
| lead-scoring | Score fit and intent against a rubric so hot leads reach a human today. | 3 |
Limits and honesty
- 1,000 decisions a month. The counter resets on the calendar month and there is no paid tier yet — if you hit the ceiling and want more, say so, because that is the signal worth having.
- The key is shown once. Only its hash is stored, so a leaked database does not hand anyone a working set — and nobody here can read your key back to you.
- No uptime promise. This runs on one site's budget. Do not put it on a path where an outage matters.
- Your text is not stored. It is forwarded to the model and the response is returned. We keep a count of decisions against your key and nothing else.