Inspiration

We all have our own aesthetics, and they're often reflected in our Pinterest boards and Instagram feeds. However, finding clothes that actually match those aesthetics can be difficult, especially when shopping across different brands. You can't exactly describe a specific vibe in words, and you can't exactly filter a clothing website by one either.

Instead of trying to describe your style to a search engine, we wanted to find a way to show it exactly what you like.

That's why we built Fitspo.

What it does

Fitspo is an AI-powered Chrome extension that transforms fashion inspiration into personalized shopping recommendations across multiple websites.

Users start by uploading inspiration photos, which Fitspo analyzes using a vision-capable AI model to identify defining style attributes, such as colors, silhouettes, textures, and overall aesthetics. These attributes form an editable style profile, allowing users to refine exactly what they're looking for.

When users visit an online clothing retailer, Fitspo analyzes the products on the page and compares them against their style profile, using AI to evaluate how closely each product matches their desired aesthetic.

Fitspo is designed to work across different shopping websites, bringing that same personalized matching experience wherever you shop.

How we built it

We built Fitspo around three main components:

  1. Visual Style Extraction: Users upload inspiration images, which are analyzed by a vision-capable large language model. The model identifies defining fashion attributes and translates them into a structured, editable style profile.

  2. Cross-Site Product Extraction: Our Chrome extension reads product information from shopping websites, including product titles, descriptions, and other relevant metadata. We designed this approach to work across different retailers rather than relying on a single website.

  3. AI-Powered Aesthetic Matching: Finally, the extracted product information and the user's style preferences are evaluated by an AI model to determine their aesthetic similarity. Fitspo uses these similarity scores to identify and recommend products that best match the user's inspiration.

Together, these components allow Fitspo to connect visual inspiration with real products across online shopping websites.

Challenges we ran into

One of our biggest challenges was translating something as subjective as personal style into information that an AI system could consistently understand and use. Fashion aesthetics aren't always defined by specific clothing items, but rather by the overall combination of colors, textures, silhouettes, and styles.

Another challenge was supporting multiple shopping websites. Since different retailers structure their product pages differently, we needed an extraction approach flexible enough to identify useful product information across a variety of layouts.

Accomplishments that we're proud of

We're especially proud of connecting visual inspiration to the actual shopping experience. Instead of requiring users to manually search for specific clothing items, Fitspo interprets the aesthetic they're going for and uses that information to identify relevant products.

We're also proud of designing Fitspo to work across multiple shopping websites, allowing users to discover clothes that match their personal style without being limited to a single retailer.

What we learned

We learned that understanding personal preferences is much more complicated than simply recognizing objects in an image. Two outfits might feature similar clothing items but represent completely different aesthetics, and capturing those differences requires a deeper understanding of personal style.

Another interesting thing we learned was how AI can help people express preferences that are difficult to put into words. When it comes to fashion, a picture really is worth a thousand keywords.

What's next for Fitspo

We see Fitspo becoming a personalized shopping companion that follows users across the web. In the future, we'd love to expand compatibility with more retailers, improve the accuracy of aesthetic matching, and introduce additional features like personalized outfit recommendations and cross-site wishlists.

Ultimately, we want to make finding clothes that match your personal style as easy as finding the inspiration itself.

Built With

Share this project:

Updates

Submission history