Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

BizzJev

Experimental lab for semantic business gates: each gate is an independent Jev Noul that code maps to NO, REVIEW or YES by threshold.

Projectprototype
View on GitHub ↗

Overview

BizzJev is a small experimental project for exploring semantic gates and decision systems with Jev. Its Semantic Operations Lab sends customer text plus all active gate definitions in one request; each gate is an independent Noul question, so several signals can be present at once with no winner-takes-all classification. BizzJev's own code then maps each probability to NO, REVIEW or YES using configurable business thresholds, and the README explains that a Noul near 0.5 means uncertainty, not medium severity.

A Decision Pipeline tab combines all three primitives in one request: a Choice for which team owns handling, a Score for urgency as the consequence of waiting (levels 0 to 3) and further questions. It defaults to TypeSafe Jev (jev-1.13.0) but can point at a compatible Kev endpoint. Live analysis needs your own key, configured with a Windows batch script, and sends submitted text to TypeSafe.

What the decision model decides

TypeQuestionAnswers
Yes / noIndependent gate per business signal (e.g. does the customer ask to cancel?)yes · no
ChoiceWhich team owns initial handling?—
ScoreHow urgent is it, as the consequence of waiting?0 · 1 · 2 · 3

How it works

  1. Submit customer text
  2. Send all active gates as Noul questions in one request
  3. Map each probability to NO, REVIEW or YES
  4. Optionally run the Choice/Score decision pipeline

Strengths

  • Multi-signal gates without forced single labels
  • Thresholds in code
  • Swappable backend (Jev or Kev)

Caveats

  • Experimental
  • Windows-oriented setup scripts

Getting started

Run SET_API_KEY.bat with your TypeSafe key, then start the lab locally.

At a glance

Type
Project
Status
prototype
Runtime
local web app
Has code
yes

Use cases and integrations

How curated lists describe it

  • The system detects all of them at once. Several signals may be present simultaneously; there is no winner-takes-all classification.

    logicrw/awesome-jev-projects

Frequently asked questions

Can several gates fire at once?

Yes, each gate is an independent Noul.

How are thresholds applied?

BizzJev's code maps probabilities to NO, REVIEW or YES with configurable thresholds.

Can I use a local model?

Yes, set SYSTEMONE_PROVIDER=kev with a compatible endpoint.

Related projects

  • Support-Ticket-Ai-Agent-with-RAG-Guardrails-and-Jev

    LangGraph student-support agent with PII masking and hybrid RAG, where Jev verifies drafts before auto-reply or human review.

    ProjectprototypePython
  • hush

    GitHub Action that triages issues with Jev, applying label, spam, needs-info and duplicate tags only above set thresholds.

    ProjectbetaTypeScript
  • judgekit

    One YAML task to classify, route, score or verify text with rules, Jev, NanoJev or an OpenAI-compatible model, with cost records.

    CLIbetaPython
  • typeful-triage

    Multiplayer GitHub triage dashboard where Jev answers kind, severity, urgency, duplicate and next step per issue.

    Appbeta
  • decisionbridge

    Adapter that turns existing LLMs into Jev-style decision functions with option scores, optional calibration and review thresholds.

    ProjectprototypePython
  • Jev Decision Gate

    CLI, MCP server and skill for issue triage where Jev predicts issue kind, affected module and whether information is sufficient.

    MCP serverprototypePython