Docs
OfficialJev Practices & Patterns
Pattern: Speculative Fan-Out
Ask everything the system might need in one request, then let code throw away what it does not use. The ticket-triage example sends a category plus four speculative questions in parallel.
- Format
- Docs
- Reading time
- 2 min
- Published by
- TypeSafe AI
- Last verified
- Sep 2026
Highlights
- All questions are evaluated in parallel, so adding questions usually has little effect on response time.
- One request carries a Choice category plus four speculative Score and Noul judgments.
- The example escalates bugs at severity above 1.5 with repro above 0.6 and flags refunds above 0.7.
- Speculative answers that do not match the branch are simply ignored.
Quickstart
python
if category.choice == "bug_report":
if bug_severity.score > 1.5 and bug_repro.noul > 0.6:
escalate_to_engineering(ticket_id, severity="high")
else:
add_to_bug_backlog(ticket_id)
elif category.choice == "billing":
if refund.noul > 0.7:
route_to_billing_with_flag(ticket_id, refund_likely=True)
else:
route_to_billing(ticket_id)Watch out
Every speculative question adds input tokens on every call, so fan out on questions that several branches can reuse.
Questions about this entry
Short answers drawn from the linked project or page.
Usually very little: questions are evaluated in parallel, so response time barely changes. The cost is input tokens for every speculative question on every call, so fan out on questions several branches can reuse rather than on everything a workflow might ask.
It sends a Choice category plus four speculative Score and Noul judgments in one request, then escalates bugs at severity above 1.5 with reproduction above 0.6 and flags refunds above 0.7 — simply ignoring answers the taken branch never uses.
You include questions that may not matter to the final decision, such as bug severity or refund likelihood, because extra questions cost little, and your code ignores the ones that turn out to be irrelevant.
After the category Choice, code only reads bug severity and reproducibility for a bug report, the refund Noul for billing, and treats frustration as useful regardless of category.
All questions are evaluated in parallel, so adding more usually has little effect on response time, and speculative questions save a round trip when they matter instead of asking category first then severity.
Escalate a bug when severity scores above 1.5 and reproducible steps exceed 0.6, flag a likely refund above 0.7, and flag frustration above 1.5 regardless of category.
A category Choice, a bug severity Score, and Noul questions for reproducible steps and refund requested, all sent against the same ticket state.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.
FacebookClinicnotes
X: OpenRouter Decision Model Rankings
X: herdr + jev, an agent router
X: A group chat bot with decision room routing
More Jev resources like this
Classify intent with a Choice and complexity with a Score, then send each branch to deterministic code, a specialist model, or a person, with a 0.5 intent-confidence floor.
Jev Practices & PatternsDocs#official#patterns#routing
Official
Use the answer to decide what to do and confidence to decide whether to act: the voice-banking example routes below 0.6 to a human and needs 0.85 or more to auto-approve a transfer.
Jev Practices & PatternsDocs#official#patterns#confidence
Official
Architectural patterns for System One systems: speculative fan-out, confidence-gated routing, composite scoring, and intent routing.
Jev Practices & PatternsDocs#official#patterns#architecture
Official
An independent calibration study of Jev against a frontier fallback on Banking77 and Web of Science, packaged as the janus-decide tool that measures a routing threshold on your own data instead of shipping a default one.
Jev Practices & Patterns#community#python#routing
Community
FirasSX914/JanusPython
FirasSX914/JanusPythonAn open lab for using Jev correctly next to Claude Code: a patched jev-router with per-turn model routing, a health-gated fallback with alerts, and write-ups of two silent bugs and of where tokens actually go.
Jev Practices & Patterns#community#catalog-import#logicrw
Community
pasblinn/jev-labTypeScript
pasblinn/jev-labTypeScriptJevCascade allocates inference-time compute for prefill-only decision models using readout stability: when only the candidate menu changes, post-intervention accuracy is already determined by the cached first-pass distribution. Menu-only interventions are predicted to within 4.2 points across six model families.
Jev Practices & PatternsArticle#community#research#arxiv
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.