Jev ecosystem directory — model live 2 days ago

Hunt the frontier of
machine-native software.

JevHunt indexes every app, playbook and tool built on Jev — the System One model from TypeSafe AI that returns typed decisions with calibrated confidence. Discover what the ecosystem is building, then ship your own.

/
Browse the directory Official Jev docs ↗
jev · decision.py ~120ms 
█
intent→"refund" p=0.94 · needs_human→false p=0.88
0
Apps indexed
0
Categories
0
Faster than chat
0
Days since launch
01

What is Jev?

A new class of model that speaks in decisions, not paragraphs.

Decisions, not strings

Jev returns typed outputs your software can act on directly — enums, booleans, scored integers. No parsing prose, no fragile regex.

Calibrated confidence

Every decision ships with an estimate of how sure the model is. Act autonomously above your threshold, escalate to a human below it.

Zero hallucinations

Jev doesn't write prose, so there's nothing to hallucinate. Outputs stay inside the schema you declared at call time.

More like code

Reliable, fast, self-consistent and type-safe — built with a new architecture, sampler and a training algorithm called RLCD.

Workflow intelligence vs. cost

Intelligence per dollar, off the charts.

193.6x
Faster on System One tasks
444.6x
Cheaper per workflow
$42
Per billion input tokens
238x
Lower input price than chat LLMs
02

The directory

Apps, recipes and open-source tools built on Jev. Filter, search, ship.

42 apps sorted by signal
03

Browse by use case

Where machine-native decisions create the most leverage today.

04

Playbooks

Copy-paste patterns for the modes builders use most.

classify.py
05

Two days in

Jev is brand new. Here's what has happened — and what's next.

◢ Open submission

Built something on Jev?
Get it hunted.

The catalog grows with the community. Submit an app, a mode recipe, a typed schema or an open-source plugin and we'll feature it in the directory.