The problem
Recommending what somebody should wear is a harder consumer problem than it sounds. Taste is personal, the catalog moves, and the product has to be worth opening again on a Tuesday, not just on the day it was downloaded.
Tailorie put AI recommendations and a social layer together, so a person could get suggestions that learned from them and share what they found with people whose taste they trusted.
What we built
- One backend serving every client, so a user, a session and a recommendation meant the same thing everywhere.
- Two web applications: one for the public and a separate one for customers. The Angular client alone carries 553 commits in five months, which is the pace the program actually ran at.
- A native iPhone client, shipped to the Apple App Store.
- A native Android client, built to the same backend.
Frank led design and development across all of it: front end, backend, web and both phones.
Where programs like this come apart, and how we kept it together
Four surfaces against one backend is where most consumer programs of this size fail. The web applications and the phones have to agree about what a user is, what a session is and what happens when the network is not there. That agreement has to be designed once, up front, rather than negotiated three times as each client team discovers it.
On Tailorie that agreement lived in the backend and the clients followed it, which is why one team could carry web, iPhone and Android at once.
- Angular
- iOS
- Android
- Apple App Store
- Recommendation engine
- Social features
- REST API
What we can start on
A consumer application across both stores from one team. A recommendation or personalization feature with a real service behind it instead of a demo.