Phia-Stylist

Telegram conversation with Phia AI stylist

Inspiration

This app was inspired by my friend Karla Alas who is a model and an upcoming stylist. She frequently sends me pintrest boards containing styles she picks out for me. However, I still have to find a retailer to buy the clothes, and have no information on the prices, only on what the styles look like.

What it does

The app provides a way for stylists to showcase their portfolios to perspective clients, and to manage relationships with them. Stylists can make designs and send them to customers which they can then buy through Phia. The stylist only has to worry about how to style, and Phia will automatically find the best price.

Additionally, the customer can choose to share information such as their calendars and emails. AI agents will automatically parse these data and mine relevant info, such as shopping receipts to populate one's closet, letting the stylist know what the customer already has. The stylist can also see upcoming events that the client might need to be styled for, increasing the personalization of their designs.

How we built it

We built a simple web app with a React Frontend and a Python FastAPI backend. We used Microsoft Foundry as the model provider for GPT 5.4, and Microsoft Agent Framework as the agent runtime. The application was containerized with Docker and then deployed on Azure Container Apps.

Accomplishments that we're proud of

In only one day, we proved that our solution can increase personalization for shoppers and save them money with Phia, and help automate work for stylists and enable them to better scale their businesses.

What we learned

We struggled a lot with ideation and eventually arrived at our idea by clearly defining the business problem that we wanted to solve. That problem is principally being that shoppers want personalized recommendations, and that stylists want to spend more time being creative and less time doing things like responding to messages.

The technology solution is just the "how", but through this project, we learned that one really needs an ironclad "why".

What's next for Phia-Stylist

Increase the integration with shopping platforms & Phia extension to enable "one-click-buy"
spin up more agents to automate research of styles
allow agents to factor in a stylists "design philosophy" and a customer's stated preferences
build CRM for stylists, automate scheduling of meetings