VisualFlow

👋 Hey Product Hunt! I’m Anggi, the maker of VisualFlow. We built this after repeatedly running into the same frustration with LLM prompts: they don’t usually break all at once — they slowly become fragile. A small change here, a tweak there, and suddenly the output shifts with no clear reason why. VisualFlow is our attempt to make prompt changes visible and safe: seeing which part affected the result, testing one change at a time, and going back when something breaks. This is an early launch and we’re mainly here to learn. I’d love to hear how you handle prompt iteration today, what feels painful, or where this approach falls short. Thanks for checking it out 🙏