gpt-vs-jev

Learning demo that sends the same input to GPT for text and to Jev for three Noul probabilities, side by side.

DemoprototypeMIT

Overview

GPT vs JEV is an arena-style learning demo. The same input goes to two different kinds of model and the outputs appear side by side: GPT generates short natural-language text, and Jev returns structured Noul answers, each the probability that the answer is yes. The models are not chained, and the author stresses it is not a claim that one is better; they do different jobs.

Jev is asked three questions: does this convey urgency, should a human support agent review this, and is this about a financial transaction. The README explains that a Noul near 0.5 means yes and no are similarly likely rather than medium intensity, and that there is no separate confidence field. Example buttons replay saved answers without API calls, and live custom runs are limited to two per visitor per day. Keys stay server-side in a Next.js route handler.

What the decision model decides

Type	Question	Answers
Yes / no	is_urgent: Does this convey urgency?	yes · no
Yes / no	needs_human: Should a human support agent review this?	yes · no
Yes / no	is_financial: Is this about a financial transaction?	yes · no

How it works

User enters text or picks an example
Server calls GPT and Jev in parallel
Display GPT text beside Jev probabilities
Explain how to read Noul values

Strengths

Clear teaching of what a Noul means
Saved examples need no API calls
Keys never reach the browser

Caveats

Learning demo only
Live runs rate-limited

Getting started

cp .env.example .env.local with OPENAI_API_KEY and TYPESAFE_API_KEY, npm install, npm run dev; live at gptvsjev.vercel.app.

At a glance

Type: Demo
Status: prototype
Language: TypeScript
Runtime: Next.js web app
Has code: yes
License: MIT

Use cases and integrations

Infra / SDKs / Integrations537

How curated lists describe it

Compare GPT generated language with JEV structured Noul decisions on the same…
wh000wh000/awesome-jev-live
Compare GPT generated language with JEV structured Noul decisions on the same input.
hellogumbo/awesome-jev

Frequently asked questions

Does it claim Jev is better than GPT?

No, it shows they do different jobs.

What does a Noul of 0.5 mean?

Yes and no are similarly likely, not medium intensity.

Can I try it without keys?

Yes, example buttons replay saved answers.

Related projects

Building with Jev★ 155
Agent skill for writing and improving programs that call Jev: question design, state structure, thresholds and diagnosing wrong answers.
Agent skillproduction
typesafe.pro
Server behind api.typesafe.pro, a free anonymous System One-compatible endpoint that forwards to TypeSafe on the operator's key.
Serviceproduction
jev-agent-toolkit
Portable Agent Skill teaching coding agents to use Jev as a bounded judgment layer, plus an optional MCP bridge.
MCP serverbetaTypeScript
jev-anything
Independent agent skill that makes a coding agent design, build, test and tune a bounded Jev decision layer with a contract first.
Agent skillbeta
jev-architect
Agent skill that guides you in finding, designing and testing Jev decision loops, including when to keep logic in code.
Agent skillbeta
jev-builder
Browser form for building Jev requests from pasted text without hand-writing JSON, then rerunning to see how steady the answers are.
ProjectbetaJavaScript