Before Kind Science greenlights a production budget for a brand-new SKU, I use AI-assisted concepting to pressure-test a dozen creative territories at once, the kind of creative diversity Meta's own retrieval systems now reward, so the shoot we eventually book is aimed at the direction that's actually going to win.
The target customer isn't staying home to manage a skin condition. She's on the trail, at the gym, out with her kids. The entire concepting exercise below works backward from that one insight: redness is a problem she wants solved fast, not a lifestyle she's willing to shrink around.
A new addition to the Kind Science line, built around a "MicroCalm" active blend. With no existing creative library to pull from, the launch needed a wide spread of concept directions, and fast, to see which visual language and message would actually earn attention before committing to a full shoot.
A quieter, more luxurious register: the product itself as the hero, shot with the kind of warm, tactile lighting usually reserved for higher-end skincare.
The mirror-selfie, outfit-check format people already trust in their own feed: a same-person before/after that reads as a friend's post, not an ad.
Split-face macro beauty photography across skin tones and ages, close enough to feel credible, framed gently enough to stay flattering rather than clinical.
A "person on the street" register: blind comparisons and reaction interviews, built for a platform where authenticity outperforms polish.
The "why it works" layer: high-quality ingredient photography for the audience that wants the formula, not just the promise.
Two further directions worth testing: a raw, behind-the-scenes manufacturing hook, and a clinical "science you can trust" register.
These are AI-assisted creative concepts used to pitch direction before production, not finished, shot campaign assets. That's the point: exploring this many territories this fast would not have been realistic the old way.
Twelve directions pulled from across this launch: hand modeling, UGC, before/after, street interviews, ingredients, and raw brand exploration, all built to give the algorithm and the audience genuinely different things to react to, not the same ad worn twelve different ways.
