Interim

Inspiration

When your body changes, getting dressed can stop feeling like yours. It happens with illness and its care, hair loss, surgery, weight gain or loss, a new baby, midlife, going grey, or growing out your hair or a beard. Shops show clothes on other people. Photos of outfits you loved can be hard to look at. You still know your style. What you lack is a way to see it on you, now, at your own pace.

I built Interim first for people going through those changes.

What it does

Here is the style I am going for. Here is me now.

You add up to three outfits you have worn or admire, plus a photo of you today. YouCam's AI clothes try-on shows how each outfit could look on you. Compare before and after in one tap, and save the looks you like.

• Your taste stays at the centre. Previews are ideas, never a promise of size or fit, and the app says so on every result.
• A private place to reflect. Write what you want to, add optional words for the day, and there is always a link to someone to talk to.
• Check-in profiles follow a change over time. YouCam's hair, face and skin analysis and hairstyle previews feed a Stage Card with a companion note in a tone you choose.
• Ten languages: English, French, Spanish, Italian, Portuguese, Swahili, Hausa, Yoruba, Igbo and Pidgin.
• Yours to export or delete, at any time.

How I built it

YouCam APIs

• AI Clothes Virtual Try-On (cloth_v4, garment_category: full_body): one call carries a whole outfit onto the person's photo. This is the core of style previews.
• Hair, face and skin analysis (hair length, density, type, face shape, skin tone) and hairstyle, colour and beard previews power the check-in profiles.

Agentic workflow: an agent plans which YouCam calls to make and explains the result in the person's language. Consent, eligibility and budgets are plain code that the model cannot override, and the model never sees a photo.

Stack: Next.js (PWA), NestJS API and worker, PostgreSQL (Neon), Redis and BullMQ, Cloudflare R2 for private photo storage, zod validation at every boundary, TypeScript throughout.

Challenges I ran into

• Making previews feel respectful. I kept them as ideas and never claimed fit or size. I also avoided any "ideal body" framing.
• Keeping the product safe around sensitive situations: consent before any photo leaves, a short photo lifetime, and everything exportable and deletable.
• Making a paid API demo-able without spending units. I built record and replay, so the live site serves real recorded YouCam results and a judge cannot run up cost.
• Sharing one storage bucket between development and production without either deleting the other's files. A key prefix solved it.

Accomplishments that I'm proud of

• A complete, deployed product: sign-in with email or Google, uploads, previews, saving, reflections, check-ins, export and delete.
• Plain language and ten languages, with no accessibility violations found by automated checks on the screens I tested.
• Safety and spending rules enforced in code, not left to a model.

What I learned

• Whole-outfit try-on works best when the request is simple, so the app asks for one photo of you and one of the outfit.
• People in a body change need control and calm more than features. Everything optional is optional.
• Honest limits make the product more trustworthy, not less.

What's next for Interim

• Previews for single garments, and shopping links so a preview you love can become something you can buy.
• Talking with the people Interim is for, such as patient communities and support groups, to shape what comes next.
• Native-speaker review of every language and a wider set of sample looks.

Try it

Live: https://interim.afolarin.cv. Open /demo without signing in, or sign up with email or Google. On the style screen, choose a sample set, then Show me. Code and run instructions: https://github.com/Folareen/Interim

Built With

• bullmq
• cloudflare-r2
• drizzle-orm
• gemini
• neon
• nestjs
• next.js
• node.js
• pm2
• postgresql
• react
• redis
• typescript
• youcam-api
• zod