Which team should handle this ticket?
- billing 100%
- technical 0%
- sales 0%
- account 0%
A practical guide · Playground to API
Turn a piece of text into a decision your app can use. Start with one support ticket, ask focused questions, then read the labels, probabilities and scores.
This guide uses Jev on Jev AI. New to the model? Start here.
“Our payouts have failed every day since Monday… If this is not fixed today we will have to move to another provider.”
Use an example to learn the controls, then replace the text with your own.
jev-latest.The playground allows up to 8,000 characters and 8 questions per request. API limits differ; see the API reference.
Hi, this is the third time I am writing. Our payouts have failed every day since Monday and my team cannot pay suppliers. I already re-entered the bank details twice. If this is not fixed today we will have to move to another provider.
This is the shared input for all four questions in the example.Switch between three questions about the same ticket. The output shown is a saved recording, not a live call.
Define distinct options and describe what each one means. Read the chosen label and the full probability distribution.
billingPayments, payouts, invoices and refundstechnicalBugs, outages, integrations and API errorssalesPricing, upgrades and new accountsaccountLogin, profile and permissionsWhich team should handle this ticket?
answers.team.choicebillingThis interactive example only replays recorded data. It does not call the API or use your balance.
Editing an example changes the request. Run it again to get an answer for the new input.
Live runs use your available balance. Review current pricing and account usage before running large batches.
Use choice as the selected label. Inspect probabilities to see whether alternatives are close.
noul: 0.96 means an estimated 96% probability of yes. Choose a threshold based on the cost of mistakes in your workflow.
Match the numeric score to its legend. A 3 on this four-level scale is different from 3 on a five-star scale.
Keep your state and questions. Send them from your server with a Jev AI API key.
JEV_AI_API_KEY on your serverIn the playground, API code generates cURL, JavaScript or Python from your current input. The example below sends the complete support-triage case. Running it makes a live, metered API call; the tutorial itself does not.
curl --fail-with-body https://jev-ai.pro/api/v1/systemone \
-H "Authorization: Bearer $JEV_AI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "jev-latest",
"state": "Hi, this is the third time I am writing. Our payouts have failed every day since Monday and my team cannot pay suppliers. I already re-entered the bank details twice. If this is not fixed today we will have to move to another provider.",
"questions": {
"team": {
"type": "choice",
"instructions": "Which team should handle this ticket?",
"criteria": {
"billing": "Payments, payouts, invoices and refunds",
"technical": "Bugs, outages, integrations and API errors",
"sales": "Pricing, upgrades and new accounts",
"account": "Login, profile and permissions"
}
},
"urgent": {
"type": "noul",
"instructions": "Does the customer need a response today?"
},
"frustration": {
"type": "score",
"instructions": "How frustrated is the customer?",
"criteria": [
"Calm",
"Mildly annoyed",
"Frustrated",
"Angry and ready to churn"
]
},
"churn_risk": {
"type": "noul",
"instructions": "Does the customer threaten to leave or switch providers?"
}
}
}'Set the environment variable before running the example. Keep the key on your server. Jev AI keys authenticate this site's endpoint; an official TypeSafe endpoint requires its own key.
Check the request and account state before repeating a call.
| What you see | What to check |
|---|---|
| 401 · Authentication | Send a valid Jev AI key in the Bearer header and confirm it has not been revoked. |
| 402 · Balance | Check your available balance and account usage on the Jev API page. |
| 422 · Validation | Check state, question types, unique question IDs, criteria and request limits. |
| 429 · Capacity or rate limit | Respect Retry-After when present and reduce concurrency. |
| 502 / 504 · Upstream failure | Check the error and recorded usage. Avoid blindly replaying a request when its outcome is uncertain. |
You can inspect the recorded examples without making a live request. Sign in to run your own text in the playground. An API call requires a Jev AI API key and an available balance.
No. This tutorial displays the support-triage recording dated September 18, 2026. Switching tabs only changes which recorded answer is shown. Run your own case in the playground for a fresh result.
This tutorial covers text decisions with Jev. OmniJev and Jev-Omni are separate multimodal projects; their model guides describe their own inputs and setup.
Score answers are probability-weighted across the ordered levels you defined. A result between levels is expected. Read the returned legend and probabilities rather than treating every score as an integer label.
Start with the support example, then change one input at a time.
This guide describes the Jev AI playground and API. For the separate multimodal projects, see OmniJev and Jev-Omni.