John RodriguesAI Design Engineer | Designing 0→1 AI native Products

Jev: The Decision Layer Boosting AI Retrieval from 72% to 95%. #Jev #AIDecisionLayer #GenerativeUI

22 Sept 2026 · 1 min

Watch on YouTube

0:42

John Rodrigues presents Jev, a decision-layer model that sits above LLMs to optimize cost and improve output quality. When applied to Human AI Studio's design system, it boosted component retrieval accuracy from 72% to 95% by automatically selecting the right UI component during generation.

Chapters

There is this new model called Jev and everybody is talking about it. Also, everybody's confused what it is. It's not another chatbot. That's why it's very confusing. Basically, a decision layer that exists on top of your LLMs that helps you optimize cost, makes decision behind so that you can have a better output. I applied this to our AI-ready design system and it took our design system from 72% retrieval to 95% retrieval because it's making the decision behind the scene which component to use when. This opens up the door for advancement in generative UI. I wrote a complete article on our Substack based on the research we did at Human AI Studio. So you can go read about it. Helps a lot with building automations and applications worth looking into.