Agent workflows · 2026-09-25
Explaining a Jev decision with a chat model
A guide to separating a typed Jev decision from user-facing explanation generated by a separate chat model.
Treat decision and explanation as two jobs
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. separating a typed decision from free-text explanation için kararın hangi alan tarafından tüketildiğini belirtin.
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
Read only typed results from Jev
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. application reading of Jev answer fields için kararın hangi alan tarafından tüketildiğini belirtin.
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
| Need | Type | Field |
|---|---|---|
| Fixed option | Choice | choice, confidence |
| Ordered level | Score | score, confidence |
| Yes/no probability | Noul | noul |
Give the chat model limited context
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. passing context to a chat model that writes an explanation için kararın hangi alan tarafından tüketildiğini belirtin.
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
Keep explanation separate from action authority
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. preventing explanatory text from starting an action için kararın hangi alan tarafından tüketildiğini belirtin.
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
Illustrative threshold example
const response = await client.post("/v1/systemone", { model: "typesafe/jev", state, questions });
const answer = response.answers.route;
if (answer.confidence < 0.80) sendToHuman(answer.choice);
else routeTicket(answer.choice);Improve wording with user feedback
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. reviewing explanations with user feedback için kararın hangi alan tarafından tüketildiğini belirtin.
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
Improve the question by measuring outcomes
The explanation model must not invent eligibility reasons, start an action, or change the supplied decision. If its call fails, show fixed short copy; if the Jev decision fails, hold the action or use human review. Measure decision correctness separately from whether users understand the explanation. separating a typed decision from free-text explanation için kararın hangi alan tarafından tüketildiğini belirtin.
Getting a decision from Jev and getting an explanation from a chat model are separate responsibilities. Jev's choice or Score field goes to the rule engine; the chat model writes user language only from permitted reason labels. A tone change therefore cannot make an action depend on free text. Sonuçları gerçek örneklerle inceleyin; istisnayı soru metni, seçenekler ve iş kuralı arasında görünür tutun.
Frequently asked questions
Does Jev generate explanation text?
No. Jev generates no free text; it returns typed answers.
Do I call the same model for an explanation?
No. When an explanation is needed, call a suitable chat model separately and provide the Jev decision as structured context.
Can an explanation change the decision?
No. An application action should depend on the Jev result and its own rules; the explanation is user communication.