How it works
+-
+
- Tap Love or Nope on 20 color swatches, one at a time. +
- A small neural net retrains in the page after every tap, and its neurons glow as they fire. +
- See a heat map of the color wheel, a 5-color palette and guesses on new swatches. Download a wallpaper PNG or copy a link; a partner who opens it and taps their own 20 sees both brains side by side with a match score. +
Why this exists
+Huebrain rides the wave of these repos, trending on GitHub this week. It does not use their code.
+-
+
- HeyPuter/everything, 2,170 stars +
- chenglou/handwired, 1,711 stars +
Pricing
+Nothing is charged today.
+Free
For anyone picking colors for personal use, like Dana and her partner, or a friend group.
+Includes 20-swatch Love/Nope training with a live neural net and visible neuron pulses, color-wheel heat map, 5-color palette, guess-and-check on new swatches, wallpaper PNG download, and a share link with side-by-side match score.
+ Use it free +Pro (early access)
For freelance designers, stagers and decorators who show clients palettes.
+Includes save and retrain brains over time, blend a couple's or team's brains into one shared palette, export to ASE, Figma and Procreate swatches, and match to the nearest real paint-brand colors.
+ Join early access +Studio (early access)
For small design or staging teams sharing client work.
+Includes everything in Pro, plus shared team brains, team sharing of saved palettes, and blending across several clients or teammates.
+ Join early access +Honest limits
+-
+
- The net is tiny (4 inputs, 8 neurons) and learns from only 20 taps, so its taste map is a rough guess, not a measurement. +
- Colors are shown on your screen. Real paint looks different in real light, and the page does not match to paint brands. +
- Nothing is saved. Reload and your brain is gone unless you copy the link or download the wallpaper. +
- The share link holds the net's 49 weights (not your taps), and anyone with it can see your taste map. It compares two brains only: theirs and yours. +
- The match score is the average agreement of the two nets' yes-scores over a grid of colors. It is a rough number, not a verdict. +