About
We built it for the shopper who cannot tell which one is hers.
Skincare shelves are crowded and the advice is thin. Claré gives beauty brands a way to look at the person in front of them, say what it sees, and recommend from their own range.
Our story
Making the answer true.
Getting a vision model to describe a face is easy. Getting it to be right is not. Early on, we watched it give lenient scores, invent oiliness on almost every photo, read stubble as blemishes and tell a 65-year-old that their skin was in trouble.
So we stopped polishing prompts and started building around the failures: evidence before numbers, a validator that drops what it cannot justify, a score computed by a fixed function and a recommendation engine that treats safety as a wall.
Claré is on its third version and is live. Versions one and two came before it, and we learned something from each.
Principles
What we will not do.
Guess
If a photo cannot support a claim, we drop the claim. Oiliness under flash is the classic case.
Diagnose
Claré is cosmetic analysis. It never diagnoses and never uses medical language.
Keep your face
The photo exists for one analysis and is then gone. No disk, no database, no log.
Estimate your age
Age is an optional dropdown. We never infer it from a face, and it changes the framing, not the score.
Sell against you
Every recommendation comes from your catalog. We have no products of our own to push.
Hide the limits
We publish what is not solved yet, from our small fitting set to the catalog import we have not built.
Ready when you are
Let’s look at your shelves.
Book a demo and we will walk through a scan, the checks and a routine built from a sample catalog, then talk about how it fits your brand.