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

typesafe-showcase

Next.js site with six live Jev demos, each compared side by side with Gemini or Claude on latency, cost and agreement.

Demobeta
View on GitHub ↗Live demo ↗

Overview

typesafe-showcase is a Next.js site deployed on Vercel with six live demos of Jev, each pairing a real interaction with a real judgment. Visitors paste their own TypeSafe key, stored in browser localStorage and passed to server actions per request, so there is no shared key to spend. An optional comparison picker runs the same judgment against Gemini 3.5 Flash-Lite or Claude Haiku 4.5 using the visitor's key, showing real latency and cost.

Demos include judging 16 tickets for billing sequentially versus concurrently, a three-way citation checker (supports, contradicts, says_nothing), and a four-level tone escalation ranking where the README notes Gemini and Jev agree only about 60 percent of the time. It is the web counterpart of the same author's use-case scripts.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this ticket about billing?yes · no
ChoiceDoes the cited policy support, contradict, or say nothing about this claim?supports · contradicts · says_nothing
ScoreWhere does this message fall on a four-level escalation rubric?—

How it works

  1. Paste a TypeSafe key (and optionally a comparison model key).
  2. Pick a demo.
  3. Jev and the comparison model judge the same inputs.
  4. View results, latency, cost and agreement.

Strengths

  • Side-by-side comparison with general LLMs.
  • No shared server key.
  • Live hosted demos.

Caveats

  • Requires your own keys.
  • Small demo datasets.

Getting started

Visit https://typesafe-showcase.vercel.app and paste your TypeSafe key.

At a glance

Type
Demo
Status
beta
Language
TypeScript
Runtime
Next.js web app
Has code
yes
Stars
1

Use cases and integrations

How curated lists describe it

  • Next.js UI showing off TypeSafe's System One model (Jev) — parallel Noul judgments and a Choice-based citation checker, deployable to Vercel

    wh000wh000/awesome-jev-live

  • Next.js UI showing off TypeSafe's System One model (Jev) — parallel Noul judgments and a Choice-based citation checker, deployable to Vercel.

    hellogumbo/awesome-jev

Frequently asked questions

Whose key is used?

Your own, stored in your browser and passed per request.

Which models can I compare against?

Gemini 3.5 Flash-Lite or Claude Haiku 4.5.

Where do Jev and the LLM disagree?

Most on the tone escalation demo, at about 60 percent agreement.

Related projects

  • warped-sys1-lab

    Local SvelteKit workbench and CLI for organizing, editing, running and comparing Jev experiments defined as JSON.

    CLIbetaTypeScript
  • JudgeJev

    Learning lab for DeepEval with Jev as the judge: replayed evaluations, adjustable decision rules, release gates and drift alerts.

    ProjectprototypePython
  • OpenJev (GPT-AGI)

    Jev-compatible open System One engine with a Claude Code-style terminal REPL showing live probabilities, plus an experiment harness.

    ProjectprototypePython
  • jev-behavior-study

    Independent field guide to Jev 1.13.0 behaviour across 11,621 requests: framing sensitivity, Snake and 3D city driving experiments.

    ProjectresearchPython, TypeScript
  • jev-report

    Independent 52-page Chinese report on Jev that checks vendor claims and reproduces 50 Chinese tests, with traceable data.

    ProjectresearchPython
  • jev-playground (Ahmar004)

    Interactive teaching site where Jev races frontier LLMs in games, an arena, a sandbox and quizzes, with a keyless replay mode.

    AppproductionTypeScript