Triage failure modes with Jev
Incoming tickets need a queue, a temperature reading, and a sense of time pressure — then code routes. Jev judges; it does not write the reply.
This unofficial page is the failure modes slice of the support triage pack. Intent: apply the Jev (TypeSafe System One) decision model to support triage failure modes. Primary search language: Triage Jev failure modes. Confirm patterns on docs.typesafe.ai. This site does not sell, issue, or proxy TypeSafe keys. Use a credential you already have from the console or a documented gateway.
Independent angle (cover ≠ clone): Full triage pack (contracts → thresholds → handoff → audit) beats a single support-triage-choice recipe. We do not clone rival recipe IA.
Triage use-case context
Support triage breaks in product-specific ways. This page lists those modes so you can write tests — not a generic “AI can be wrong” essay, and not a rival limitations-page clone.
Hub: Support triage hub. Compare, when the other tool is the real job: keyword triage.
Failure Modes inputs
Many failures start as contract violations (distractors, missing support ticket text). Canonical shape:
{
"ticket": { "id": "T-1042", "body": "Card charged twice after the outage.", "channel": "email" },
"customer": { "plan": "pro", "open_incidents": 1 },
"policy": { "p1": "P1 = safety issue or complete outage" }
}
Decision signals and actions
- Keyword-only queues miss sarcasm and multi-intent tickets — that is why Jev sits in front, not as a regex replacement for IDs.
jev-1.13loses accuracy when CRM novels sit in state (jaggedness: distractors).- Treating
is_urgent.noul == 0.5as “medium priority” — it means yes/no are equally likely. - A 200-leaf product tree as one Choice — use a cascade with confidence abort instead.
- Asking Jev to draft the customer reply (generation is the wrong job).
HTTP vs application:
| You see | Class | Triage move |
|---|---|---|
| 401 / 422 / 429 / 529 | Documented HTTP | Fix key/body or back off — errors |
| 200 + flat confidence or Noul ≈ 0.5 | Low confidence | Hold; do not page on-call or issue a refund flag |
| Empty gather | Missing evidence | Skip Jev or ask “is enough information present?” |
Do not treat a Noul of 0.5 as a “medium” support triage score — it means yes and no are equally likely. Conjunctions stay in your code.
Guardrails and escalation
Fail closed: do not page on-call or issue a refund flag. Schema-safe answers are not factual correctness. TypeSafe’s confidence-gated examples use a lower bar for recoverable reads than for irreversible actions. Those numbers are illustrations. For support triage, treat page_oncall as the high bar (paging on-call or issuing a refund flag). Tune on labels — see offline evaluation.
Evaluation and rollout notes
Your canary set should include each bullet above.
- Auto-route error rate at your floor (not a jev.pro leaderboard)
- False pages / missed P1s — tune
URGENT_TAUand the technical conjunction separately - Handoff rate: if everything pages, the floor is theater
Pin jev-1.13.0 (the versioned id) after you fit thresholds. jev-latest and the marketing line jev-1.13 can move. Log the response model. TypeSafe’s published list price for jev-1.13 is $0.042 per million input tokens (vendor claim — confirm on the models page); output tokens are free on that same page. Unused distractors still bill as input.
Official Python and JavaScript SDKs read TYPESAFE_API_KEY and retry documented 429/529. This site does not sell, issue, or proxy TypeSafe keys. Use a credential you already have from the console or a documented gateway.
Pack map
| Slice | Page |
|---|---|
| Graph and primitives | decision workflow |
What may enter state |
input contracts |
| What to gather first | evidence collection |
| Atomic rules | policy checks |
| Act / review / abstain | confidence thresholds |
| Reviewer payload | human handoff |
| What to persist | audit trail |
| How it breaks | you are here |
| Labeled replay | evaluation |
| Shadow → canary | production rollout |
FAQ
If the API returns 200, is the decision good? 200 only means the call parsed. Low confidence, Noul ≈ 0.5, or a policy miss are application failures.
Where do official weaknesses live? TypeSafe’s jev-1.13 jaggedness note — distractors, arithmetic, adversarial content. We do not invent more.
Where is the rest of the Triage pack? Start with Triage evaluation and Triage decision workflow. Cluster hub: Use cases.
Can Jev replace the support agent? No. It classifies and scores. Agents, macros, or humans act. See the support triage hub.
Should every company copy TypeSafe’s 0.5 intent floor? No. Official numbers are illustrations. Fit on offline evaluation.
What this page does not claim
- No invented triage accuracy or SLA.
- Schema-safe
department≠ the customer named the right product. - Not official TypeSafe documentation.
- Official TypeSafe status, or that jev.pro issues API keys.
- That a schema-constrained answer is automatically factually correct.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai.
Primary documentation: https://docs.typesafe.ai. Hub: Use cases.
Sources
Public TypeSafe or adjacent documentation only. No private claims.