Is It Working?

Weekly YouCam Skin Analysis check-ins that tell you whether each skincare product is actually working: keep it, give it time, or drop it. Then try a matched shade on with VTO before you buy.

Comment

Inspiration

Most of us buy skincare on hope. We use a serum for weeks and still can't tell if it's working, so we keep paying for it or give up on something that just needed time. Products that don't work and wrong-shade foundation are a big part of why beauty purchases get wasted or returned. YouCam's Skin AI can measure skin objectively, so we built the app for the moment the brief describes: deciding whether a product is working.

What it does

  • Weekly guided check-in: the camera frames your face in an oval and checks the lighting first. The photo goes, server-side, to YouCam AI Skin Analysis (v2.1), which scores 8 concerns (acne, redness, age spots, wrinkles, pores, moisture, texture, radiance) and returns a detection mask for each.
  • Trends with product markers: per-concern trend lines show a marker for the day you started each product, and a compare view puts week 0 next to the latest week with the real YouCam masks.
  • Verdicts with reasons: a deterministic engine measures each product's target concerns, before and after, against your own week-to-week noise band and the product type's realistic time to work (retinoids need 8–12 weeks). It returns KEEP, GIVE IT N MORE WEEKS or DROP, with the numbers behind each one.
  • Try before you buy: for a dropped tinted moisturiser, YouCam Skin Tone Analysis reads your skin colour, the app picks the nearest shade by colour distance (ΔE), and YouCam Makeup VTO previews foundation and concealer on your own face. ## How we built it

Next.js 15 + TypeScript on Vercel, built mobile-first. Every YouCam call happens in a server route, so the API key never reaches the browser. Each one follows the same steps: upload with the File API, start an async task, then poll until it's done. Before each call the app checks the credit balance. Live calls need an access code and are rate-limited, and on low credit or an error the app shows a recorded real response instead. The verdict engine is pure and unit-tested. Your data stays on the device (localStorage).

Challenges

Weekly photos are noisy. Lighting and angle move the scores, so we guide the capture and judge every change against each user's own noise band rather than a fixed threshold. A small credit budget shaped the fallback design.

Accomplishments

It's a complete loop: capture, measure, link the change to a product, decide, then buy the matched shade. Every verdict can be explained with real numbers.

What's next

For retailers: "add the matched shade to basket" and repurchase reminders for KEEP products. Also HD skin analysis and optional consented cloud sync.

Consumer & eCommerce value

Consumers waste less money on products that don't work and return fewer wrong shades. Retailers get confident repurchases and shade-matched recommendations at the moment a customer is about to switch.

Built With

Share this project:

Updates

Submission history