User repository on GitHub
Experiments with TypeSafe/Jev semantic gates and a Semantic Operations Lab demo.
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Demos & Experiments
BizzJev
Experimental C# lab for TypeSafe/Jev semantic gates. The Semantic Operations Lab maps Noul probabilities to NO / REVIEW / YES through configurable thresholds, and the Decision Pipeline combines a Choice, a Score and a Noul in one request.
- Stars
- 1
- Language
- C#
- Reading time
- 3 min
- Published by
- Community
- Author
- havietkok-sys
- Last verified
- Oct 2026
Used inJev for Customer Support
Highlights
- Each semantic gate is an independent Noul question, so one message can trip several gates: a customer can report a billing problem and ask to cancel at once.
- The Decision Pipeline composes all three primitives in one request: a Choice for the owning team, a Score for urgency 0-3 and a Noul for cancellation intent.
- Configurable thresholds map probabilities to NO / REVIEW / YES; a Noul near 0.5 means uncertainty, not medium severity, the README stresses.
- Architecture rule: Jev owns all semantic inference while C# owns validation, deterministic policy and control flow, with no keyword rules or second models.
- Policy replay recomputes decisions from saved answers in C# with zero Jev calls; a 40-case synthetic evaluation is documented in milestone2docs.
Watch out
Grounded on the truncated README and GitHub metadata reviewed 2026-10-15; the README is cut mid-sentence inside the milestone 2 architecture rule and the repo lists no license. Live analysis needs your own TypeSafe key and sends the submitted text to TypeSafe's hosted API.
Questions about this entry
Short answers drawn from the linked project or page.
A small experimental C# project exploring TypeSafe/Jev semantic gates and semantic decision systems, built around a Semantic Operations Lab that maps Jev probabilities to NO / REVIEW / YES using configurable business thresholds.
All three primitives: Noul for yes/no gates such as cancellation intent, Choice for which department or team should handle a request, and Score for ordered levels such as urgency from 0 to 3.
It composes a Choice, a Score and a Noul in a single Jev request over one shared customer message; the typed answers feed a deterministic C# policy that proposes a team, a priority, cancellation handling or explicit human review.
No. Every proposal is a recommendation for a person, never an executed action. Jev owns all semantic inference while application code owns validation, deterministic policy and control flow, with exactly one HTTP request per run and no automatic retries.
Your own TypeSafe API key from console.typesafe.ai/keys, configured locally by running SET_API_KEY.bat; live analysis needs an internet connection and sends the submitted text to TypeSafe's hosted API. Reading the docs and saved results needs no key.
Yes. The backend defaults to Jev at api.typesafe.ai with jev-1.13.0, and SYSTEMONE_PROVIDER=kev switches to a compatible endpoint configured through SYSTEMONE_URL, SYSTEMONE_API_KEY and SYSTEMONE_MODEL. The key is sent only by the backend in the Bearer header.
The repo documents a frozen milestone 2 specification with questions, thresholds, decision rules and a 40-case synthetic evaluation under milestone2docs. The README visible in this digest is cut off mid-sentence, so later sections were not reviewed.
More Jev resources like this
A workshop of small, inspectable Python experiments for TypeSafe System One models: ten critical bays, from fraud screening to emergency dispatch, each giving Jev a realistic state while deterministic policy keeps authority over the next action.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
jgalego/jevs-garagePython
jgalego/jevs-garagePythonA no-dependency single-page classifier that unpacks a Jev call: paste an API key, edit a choice question, and see the request JSON, per-category probabilities, confidence, latency, token usage and the equivalent curl.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
q93304989-bit/jev-labHTML
q93304989-bit/jev-labHTMLA customer-service chatbot routed by Jev: one API call per turn asks every question in the decision tree up front and discards the unneeded answers — the docs' speculative fan-out — with a human-handoff Noul and a frustration Score riding along.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
minghanminghan/jev-demoTypeScript
minghanminghan/jev-demoTypeScriptVercel Labs example that routes three forms by context with Jev and AI SDK: Jev's team choice is accepted when its confidence is at least 95%, otherwise openai/gpt-6-luna-fast independently evaluates the submission and its choice becomes final.
Jev Demos & Experiments#community#catalog-import#logicrw
An autonomous ViZDoom agent that decouples Doom's simulation from Jev SystemOne inference: the game runs at 35 ticks/sec while Jev is queried asynchronously at about 10 Hz for macro goals, targets, movement and firing decisions.
Jev Demos & Experiments#community#catalog-import#logicrw
Community
amoghcreator/doom-jevPython
amoghcreator/doom-jevPythonA research pipeline by CompleteTech LLC AI Research that downloads a slice of NYC 311 reports, scores complaint descriptions with Jev, and renders severity, safety-concern, and ambiguity signals as interactive maps and print-ready PNGs on a 500-meter grid.
Jev Demos & Experiments#community#catalog-import#logicrw
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.