The Tech Behind TasteBrain

Let's Chat

Let's Chat

You're viewing in ELI5 plain english,
want to get more technical?

You're reading in ELI5 plain english,
want to get more technical?

Switch to Sorkinese™

Switch to Sorkinese™

Most systems can tell you what something is. Our model understands why people prefer it.

Taste is constantly evolving, so our system continually learns from new products, brands, and cultural trends instead of relying on outdated data.

Vision-language models are powerful, but on their own they don’t capture the subtle cues that shape taste and preference.

Instead of treating each item as a single data point, the model looks at many aspects at once—how it looks, what brand it is, how it’s made, and how it’s used.

Each person has their own TasteGenome, learned on top of the same system, without rebuilding the model every time.

Because taste is modeled explicitly, the system can answer open-ended questions like why something fits—or doesn’t fit—your preferences.

Instead of relying on basic “similar item” logic, the system understands aesthetic relationships across domains, enabling more meaningful and unexpected discovery.

Let's Demo This Thing

Let's Demo This Thing

COMMERCE IS BROKEN

LET'S FIX IT