Made with Jev
Everything here is made with Jev — the same typed, calibrated decision that returns one key, a probability for every option, and a confidence score in a single round trip. Open any of them, then build your own.
Jev is TypeSafe AI's first System One model, launched in September 2026 by Diogo Almeida — the OpenAI researcher who co-invented RLHF and InstructGPT, the methods behind ChatGPT. Instead of writing prose, Jev emits typed structured values directly: one of the classes you defined, calibrated probabilities across all of them, and a confidence value. The demos below all run through the same hosted endpoint you get with a key, so what you see is exactly what you can ship.
Why 'made with Jev' looks different
Jev does not generate text token by token. A single non-autoregressive pass produces every answer at once, so there is nothing to parse, nothing to repair, and — because the output is locked to the type you declared — nothing to hallucinate. That is what lets a tool built on Jev return a decision in roughly 70–500ms, about 40–200x faster than prompting a frontier LLM.
Tools made with Jev
Games made with Jev
The games are the same primitive turned into an opponent or a referee: every move is a typed Jev decision, so you can watch a calibrated model reason under real constraints.
Make something with Jev
Grab a jv_live_ key and hit POST https://jevtypesafeai.com/api/v1/decide (or the official POST https://api.typesafe.ai/v1/systemone). Define your options, send the state you want a decision on, and you get back a typed answer you can act on immediately. That is all any tool or game on this page does.
FAQ
What does 'made with Jev' mean?
Every tool and game here calls the Jev decide endpoint for its decisions — the same typed, calibrated API you get with a key. There is no separate SDK magic; each one sends state plus typed questions and acts on the typed answer.
Can I build the same things?
Yes. Try any demo free in the browser, then generate a jv_live_ key and point your own code at the same endpoint. The request shape is identical to what powers these pages.
Why are they so fast?
Jev answers in a single non-autoregressive pass rather than generating text token by token, so a decision returns in roughly 70–500ms — about 40–200x faster than prompting a frontier LLM.
See also: All tools · All games · What is Jev · Pricing
Make something with Jev
Try it free in the browser, then grab a jv_live_ key and ship your own.