What it is
An experimental .NET architecture for running many semantic decisions through one validated path, with a labelled 111-case benchmark comparing TypeSafe Jev against a Claude baseline and a dashboard for inspecting any single decision.
How it uses Jev
Jev answers atomic Noul, Choice and Score questions defined in versioned decision packs. One fabric batches questions sharing a state, validates the typed evidence, then a deterministic policy applies thresholds, risk gates and escalation to produce an allow, approval or deny disposition.
Primitives:choicescorenoul
Technique worth stealing
Separate raw classification, policy transformation and executable decision into three layers, and report them apart.
Try it
dotnet test JevEnterpriseDecisionFabric.sln --configuration Release
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-centricyes0.92
- Shows a System One patternyes0.73
- Handles uncertaintyno0.18
- Measuredno0.28
- Runnableno0.34
- Worth recommendingunclear0.48
- Model replicano0.05
- Problem scopescore on a 0–2 scale1.19
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 21 Sept 2026.