Super Candy Labs
Beta

AI is going to destroy the world. First, it's going to make the best candy bar.

Why two brothers started a candy lab, and what we're actually building.

We’re brothers, and we have never agreed on the best-tasting candy. Not once. So we stopped arguing and started a lab.

Here’s the plan. You scan your tongue, answer a few questions about what you actually like, and we match you with real candies you’ll love. Every scan and every answer goes into one growing taste database. We’re using that database to design Super Candy: bars built from what thousands of real palates want, not what a focus group tolerated.

One candy can’t be everyone’s favorite

So we’re building more than one. Preview Model #01 is the first, and about 4 in 5 palates qualify for it. The palates it misses are the most interesting data we have: they tell us what Model #02 should be.

Honest about the bit

The tongue scan is a ritual, not a medical instrument: it runs on your device, and adults’ scan photos go into a private research dataset you can delete any time. Your answers, your favorites and the candies you pick are the real signal. We’ll publish what we learn here, including benchmarks, the research we lean on (our white paper has the citations), and the moments we get it wrong.

AI is going to do a lot of things to the world. We’d like it to start with dessert.

ShareXLinkedInText
More from Lab Notes
  1. September 28, 2026UpdatePalate taxonomy v0.5: new type names, and how the lab classifies a palate
  2. September 28, 2026ResearchCandyBench v0.1: the first candy benchmark
  3. September 28, 2026GuideThe Halloween bowl, sorted by palate type