Jev is a new kind of model from TypeSafe AI. It takes unstructured context and returns typed, probabilistic decisions that software can act on.
Using Jev well changes how you design workflows. You have to decide which judgments belong to the model, how to frame those decisions, and when the result should trigger an action, human review, or escalation. The fastest way to develop that instinct, or understand the “Jevability” of a problem, is to build.
Join us at the PromptQL office on Tuesday, September 29, from 4–7 p.m. PT for a hands-on evening exploring what becomes possible when decisions get fast, structured, and cheap.
We’ll start with a lightning session dissecting the architecture of four or five real systems built with Jev, and show off some of the best builds from the previous session. We’ll look at patterns such as routing work, selecting from existing options, ranking evidence, verifying agent output, and turning confidence into reliable control flow.
Then we’ll hack.
Bring a workflow, product idea, or recurring judgment you want to rebuild. You’ll leave with a working prototype, a sharper sense of where Jev fits, and practical patterns you can reuse in future systems.
Schedule4:00 PM: Doors open and introductions4:15 PM: Jev architecture lightning talks5:00 PM: Hackathon6:30 PM: Demos and discussion7:00 PM: Wrap
No Jev or app-building experience is needed. Engineers, founders, product builders, and anyone curious about decision-driven automation are welcome.
This is a small, hands-on gathering hosted by PromptQL. Come ready to build. Dinner will be provided.
