Frank Rhino

Disclosure up front since this is the self-promotion forum and I'm the one building it: I'm one of the people behind Clad9, an AI wardrobe app. Posting here on purpose, not sneaking it into a thread figured the build story might be more useful to this crowd than a launch announcement.

Every "AI closet" pitch sounds identical for the first thirty seconds: photograph your clothes, get outfit suggestions. The differentiation always turns out to live one layer down, in which single signal the product picked to be good at and that's also where the category quietly fails.

Look at what's actually out there. One competitor nails color harmony genuinely good color theory, matches pieces that visually work together. Another does a chat-style interview to build a style profile. A third tracks wear history and layers in weather data so you stop repeating the same five outfits. A fourth proved the whole concept works, but shipped iOS-only. Each of these is a real, defensible product. None of them constrain on more than one axis at a time.

The reason that matters more than it sounds: color, body, and calendar aren't independent variables you can bolt together later they interact. A color pairing that's textbook-correct (complementary hues, right off the color wheel) can still clash because it ignores undertone: burgundy is a cool red, rust is a warm red, and pairing either with the wrong-temperature green looks subtly off even though the hue math checks out. Body proportion changes which of those "correct" pairings actually reads well on a given frame. And the calendar changes the formality floor underneath both an outfit that's perfectly color-matched and well-proportioned can still be the wrong choice if the day calls for an interview instead of a coffee run. Ship a product that only solves one of these and you've built a genuinely good tool that still gives wrong answers some mornings, because the other two axes are quietly overriding it.