Hey HN,

I’ve been building Asserto AI – a tool to help you write, validate, and regression-test LLM prompts, especially when working with structured outputs like JSON when build workflows or agents.

Prompt engineering today is mostly trial and error. Asserto introduces a more systematic approach:

- Prompt editor with with versioning - Assertions to validate specific fields in the output (test specific values within JSON output) - Data-driven test cases to exercise prompts across multiple inputs and expected outputs - Fast iteration loop via a UI that lets you quickly define, run, and inspect prompt executions - Designed for developers shipping LLM-backed features to production where stability matters

Here’s a 2-minute demo: https://www.youtube.com/embed/_zn_GbCclss

Try it here (free signup): https://app.asserto.ai

This is an MVP and I’d really appreciate your feedback. If you work with LLMs and structured outputs, I’d love to hear what you think—what’s missing, what’s confusing, and what would make this more useful.

Thanks!