YearlyPhotos — what I'm building instead of a deal-flow studio
Three days ago I sketched a freelancer's deal-flow studio that wired every solo-shippable Adobe API into a single workflow. I committed in writing to shipping the first slice by 2026-05-28. Then I sat down to actually spec it and realized I was building portfolio, not product.
The audience that workflow targets — freelance creatives and boutique agencies — already has best-in-class for every station. Canva for the canvas. Notion or HubSpot for the proposal. DocuSign for the signature. A re-stack on Adobe APIs is technically interesting, but it doesn't beat the existing tools individually. And Acrobat Sign requires a paid subscription baseline, which kills the free-path story for any buyer I'd actually sell to.
The same six primitives unlock a different shape. A personalized print product.
I'm calling it YearlyPhotos. The domain — yearlyphotos.com — is registered. The repo is scaffolded. V1 is the annual family photo calendar.
The buyer
Someone making a gift. The annual family photo calendar for the holidays. A baby's-first-year book. A wedding album. A travel book from the trip. They've got hundreds of photos on their phone and an idea of what the finished thing should feel like — they just don't want to design it.
This is a multi-billion-dollar category already. Shutterfly, Mixbook, Artifact Uprising, Chatbooks. The buyer is there and already paying. What changes with Adobe primitives is what the finished product can do: themed AI fills for awkward photos, generative backgrounds where the original frame is cluttered, expand a portrait to a landscape spread without cropping a face. The output crosses the line from "stitched together from my camera roll" to "something you'd actually frame."
What the product does
Upload a photo library. The app clusters by date and event, surfaces the best frames per month or per spread. The buyer picks a template — calendar grid, photo book layout, holiday card — and drops photos into slots. The system fills gaps, expands portraits to fit, generates themed backgrounds where the original photo isn't strong enough on its own.
The polished assets render into the finished product from a template. PDF preview in-browser. Buyer pays, downloads a print-ready file, or — in a later phase — orders through a print-on-demand integration.
Five Adobe APIs, doing real work
Firefly generates themed backgrounds — Halloween, Christmas, beach, autumn — and fills awkward parts of a photo: cluttered backdrop, blurry edge, missing context behind the subject in the foreground.
Photoshop API runs server-side batch processing on every uploaded photo — auto-tone, smart crop to the template slot, brand frame overlay, watermark.
Document Generation is the layout engine. A .docx template with named slots, a JSON payload with photo URLs and captions, out comes the finished calendar or book.
PDF Services exports a print-ready file in /X-1a for any print-on-demand vendor.
PDF Embed is the in-browser preview before the buyer pays.
Express add-ons and Acrobat Sign drop out. B2C consumers don't use Express, and there's no signature flow in a gift purchase.
The economics
Prototype cost is zero — free Firefly credits cover the build. Production variable cost is dominated by Firefly: a 12-month calendar runs 12–24 generations at $0.02–$0.10 each, so $0.25–$2.50 per book in raw image cost. Photoshop API and Document Generation are sub-cent per book. PDF Services has a 500-transaction free tier; production rolls into a volume contract once it earns one.
Retail target $19–49 per book. Margin is fat enough to absorb paid acquisition through Q4.
MVP cut — the calendar
V1 is the annual photo calendar. One template, one product, one funnel. Q4 has the natural launch window — by mid-October every "make Christmas gifts" search should find this product.
Out of V1: photo books, holiday cards, weddings, baby's-first-year. Out of V1: print-on-demand integration — V1 outputs a print-ready PDF the buyer takes wherever they like, which keeps cost flat and complexity low. Out of V1: mobile apps. Web-only.
In V1: upload, cluster by month, AI scene fills for awkward photos, template grid, brand frame overlay, in-browser preview, Stripe checkout, print-ready PDF download.
What's next
The spec is locked, the repo is scaffolded, and the twelve-phase build plan is on the board. Phase 0 — the runnable shell — is up next. Q4 holiday window is the launch deadline: by mid-October, every "make Christmas gifts" search should find a real product.
Project card with the full pitch: YearlyPhotos on builddaily.
The Adobe APIs earn their slot when they're earning a buyer, not a portfolio.