What it is
A reproducible demo that uses TypeSafe's Jev to triage simulated warehouse robot incidents, comparing cost and latency against a self-hosted ModernBERT model. For robotics and edge AI practitioners evaluating decision-layer costs.
How it uses Jev
Jev makes three simultaneous judgments per incident: escalate to human (yes/no), owning team (choice), and urgency (score 0–2). State is a bilingual incident template. Results are used to compute agreement with template labels and fleet-scale cost models.
Primitives:choicescore
Technique worth stealing
Using one Jev call to get multiple typed judgments, reducing cost per decision.
Try it
Run demo-a.py with OPENROUTER_API_KEY set; see REPRODUCE.md for details.
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.90
- Shows a System One patternunclear0.41
- Handles uncertaintyno0.07
- Measuredyes0.70
- Runnableno0.08
- Worth recommendingunclear0.44
- Model replicano0.05
- Problem scopescore on a 0–2 scale0.48
- About Jevyes0.99
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.