Founder
Kislyn Bejagan
Founder & Chief Executive, ExactHue. Founded the company in 2025 on the view that the beauty industry's shade problem had been misdiagnosed for a decade.
The reframing
Everyone was solving the wrong problem.
The prevailing assumption in virtual try-on was that the previews needed to look better. More realistic rendering, smoother compositing, cleverer overlays. Enormous effort went into making the picture more convincing.
The evidence pointed somewhere much less glamorous. The previews were not failing because they looked unconvincing — they were failing because nobody was modelling the light. A beautifully rendered shade computed from an unmodelled illuminant is still the wrong shade. It just arrives at the wrong answer more persuasively.
Once you take that seriously, the whole shape of the company changes. It stops being a graphics problem and becomes a measurement problem, and measurement problems have an established literature, published standards, and instruments you can check yourself against. That is a far better position to build from than a rendering arms race.
It also determined the order things got built in: a colour laboratory before a marketing team, a sampling plan before a training run, and a product that declines to answer rather than guess. Slower, and the only version that holds up when a parcel actually arrives.
“We are not trying to make the preview beautiful. We are trying to make it true. Beautiful is what happens when a customer opens the box and it matches.”
Focus areas
- Applied colorimetry and appearance modelling
- Representative dataset design across Fitzpatrick I–VI
- Commerce integration and returns economics
- Consent-first research practice and data governance
Convictions
The five that shaped the product.
These are not values-page decoration. Each one costs something concrete, which is how you can tell whether a company means them.
A number beats an adjective
“Great match” cannot be checked, argued with or improved. ΔE00 0.7 can be all three. The cost: we can be proved wrong in public, and occasionally we are.
Averages hide the failure
Accuracy gets reported per tonal band, never blended. The cost: our headline figures look less impressive than they could, because the weakest band is always visible.
Refusing to answer is a feature
A declined capture beats a confident wrong shade. The cost: demos are less magical, and we do not bill for the calls we decline.
Coverage is designed, not discovered
A balanced library comes from a sampling plan, not from whatever was easy to collect. The cost: this is the single largest line item in building the engine.
The preview has to survive the parcel
No smoothing, no brightening, no quiet lightening of the skin. A flattering preview converts marginally better and returns considerably more, and the second number is the one that decides whether a brand stays. The cost: our try-on looks less polished than competitors who are willing to flatter.
Get in touch directly.
Brands carrying a returns problem, investors looking at the category, and anyone who thinks the reasoning on this site is wrong — all welcome.