jev (zeke)
Research notes on Jev plus a live Cloudflare Worker playground that triages support messages by department, urgency and frustration.
Overview
This repository by Zeke combines readable research notes on Jev with a working demo running through Cloudflare AI Gateway. The live playground takes a pasted support message and has Jev triage it in one call by department, urgency and frustration, rendering probability bars, confidence scores and a routing decision.
The notes explain the background of TypeSafe and the name (Kahneman's System 1 and the Jevons paradox), describe Jev as a smart if statement, and walk through state and the three question types. A section on where it breaks summarises TypeSafe's jaggedness notes: no reliable arithmetic or counting, weak date and version comparison, scores that should be used to threshold or rank rather than interpolate, trouble with indirection such as double negatives, context rot from irrelevant state, not yet hardened against adversarial input, and no ability to write anything. Technical details of the example project are in a separate AGENTS.md.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which department should handle this support message? | — |
| Score | How urgent is it? | — |
| Score | How frustrated is the customer? | — |
How it works
- Paste a support message into the playground
- One Jev call via Cloudflare AI Gateway answers three questions
- Display probability bars and confidence
- Show a routing decision
Strengths
- Clear explanation of strengths and documented weaknesses
- Live demo via Cloudflare
Caveats
- Notes and demo, not a library
- Limitations summarised from vendor jaggedness notes
At a glance
- Type
- App
- Status
- prototype
- Language
- TypeScript
- Runtime
- Cloudflare Worker
- Has code
- yes
Use cases and integrations
How curated lists describe it
Research notes and an interactive Cloudflare Worker demo for Jev, TypeSafe AI's structured decision model.
hellogumbo/awesome-jev
Frequently asked questions
What does the demo do?
Triages a pasted support message by department, urgency and frustration in one Jev call.
Where does Jev struggle according to the notes?
Arithmetic, date comparison, indirection, irrelevant context and adversarial input.
How is Jev reached?
Through Cloudflare AI Gateway.
Related projects
Worked Jev examples for ticket triage and reranking, runnable through OpenRouter without an early-access key, with offline tests.
ProjectprototypePythonRunnable Jev cookbooks, a plain-English guide and sixteen interactive simulations of how a typed decision model changes architecture.
AppbetaPythonPublic Jev playground and API site that answers typed Choice, Score and yes/no questions about pasted text in about 0.5 s.
ServiceSmall Python demos of Jev: parallel Noul ticket checks with measured speedup, and a three-way Choice citation and claim checker.
AppprototypePythonTwo beginner Python apps using Jev: a message sorter and a support ticket router that send uncertain items to human review.
ProjectprototypePythonPython sandbox turning Jev issue-category and reproduction judgments into testable routing proposals, with offline replay.
ProjectprototypePython