Launching today
Fit Receipt
A private fitting agent that knows when to call JEV
6 followers
A private fitting agent that knows when to call JEV
6 followers
Try on lingerie virtually; every pick has an AI judgment receipt. Fit Receipt is a reference implementation. The agent handles chat; code protects facts and pricing. JEV (TypeSafe) is a judgment model that scores trade-offs with calibrated confidence in ~300ms—not essays. Low-confidence results stay unresolved; judgments are recorded; works without JEV. Photos stay in your browser; no server photo storage. JEV sees typed fields, never the photo. Rate-limited API. Vercel + open source.
Products used by Fit Receipt
Explore the tech stack and tools that power Fit Receipt. See what products Fit Receipt uses for development, design, marketing, analytics, and more.
LLMs 2
LLMs 2
JevFast, structured AI decisions for software automation
5.0 (3 reviews)
We didn't want one chat model to do everything. Our agent talks with the person; JEV makes the judgment. It answers closed questions with fixed choices, so it can't drift into invented claims, and every answer comes with a probability. When confidence is below 0.7, we show "trade-off still to confirm" instead of pretending to be sure. Each judgment also returns a receipt we can trace. Giving the agent JEV as a tool made our recommendations more honest, not just faster.
OpenAIAPIs and tools for building AI products
5.0 (855 reviews)
OpenAI's models shaped the thinking. One of our agents designed the questionnaire, defined what each layer should and shouldn't do, and simplified the decision logic until a person could follow it. Another ran engineering acceptance and wouldn't pass a build until it held up. They made the product clear before it was clever.
Engineering & Development 2
Engineering & Development 2
Claude CodeAnthropic’s deep-context AI coder
5.0 (718 reviews)
Claude worked like a teammate who pushes back. When I mapped our physical fitting-room flow onto the product, Claude found why a new piece was never recommended, and argued that a name with "Fit" shouldn't promise a fit we can't guarantee. It kept the person's own words out of every model call, and built each change with tests before I shipped. It helped me think, not just type.
VercelThe frontend cloud. Creators of Next.js.
5.0 (1.1K reviews)
Every branch got its own Preview URL, so a real outside tester could try each change before it shipped. His feedback shaped the launch. The AI Gateway let us try JEV as a tool for our agent before committing to it, and the runtime logs let every JEV judgment keep a receipt we can trace. For a tiny team, that meant shipping and checking our work in the same place.