Hey Product Hunt 👋 I'm Jay. I work with teams turning messy business requirements into software. When I started using TypeSafe's Jev, a model built for fast, focused judgments like "which team should handle this email?", I noticed something: the model was rarely the problem. The questions were. People would write one giant "analyse this ticket" question, or options that overlap ("Billing" vs "Payments"), or scales like "1 = low, 3 = high" that mean something different to everyone. Then they'd find out the JSON was broken when the API call failed. So I built Jev Config Generator. You: 1. Connect an AI model (a free OpenRouter model works, no card) 2. Describe your use case in plain English and add examples or documents if you have them 3. Answer a couple of questions if something's unclear 4. Review the proposed Choice / Score / Noul questions, each with the reason it exists Test them on Jev with your TypeSafe key, then copy JSON, TypeScript, Python or cURL A few things I cared about: • No account, no database. Your keys go straight to the provider for that request and aren't stored. • Nothing exports until it's valid against Jev's request format. • It explains itself. Every question shows why it was proposed and flags overlaps or vague wording. It works well for support triage, lead qualification, content moderation, contract review, AI chatbot guardrails, and anything else where someone reads something and makes a quick call. There are 20 examples on the Use cases page. I'd love your feedback, especially on use cases where the suggested questions weren't quite right. That's what I'll improve next.

Jev Config Generator

From plain-English requirement to tested Jev questions

From plain-English requirement to tested Jev questions