How we pick products

Selection process

The GPT first validates the photos, translates the user's style request into wearable outfit directions, creates one 3-look visualization, and then searches for real products that match what appears in the final image.

Product links are supporting suggestions. The style concept and visualization should come first; shopping links should not drive the creative direction.

Ranking factors

• visual similarity to the final outfit image;
• silhouette, color, texture, and garment category;
• fit intent, scale, and styling role in the outfit;
• material quality signals, construction details, and retailer credibility;
• retailer relevance for a US-first audience;
• availability signals on the retailer page;
• price and value when comparable products are available;
• return policy and shopping practicality when visible;
• avoidance of costume-like items unless the user asks for costume styling.

What we avoid

Pick My Fit should avoid counterfeit goods, unsafe products, unclear listings, phishing pages, misleading product pages, adult content, gambling, weapons, illegal goods, malware, and low-confidence marketplace listings.

Affiliate relationships

Some product links may be affiliate links. Affiliate commission does not increase the price paid by the user and should not override visual relevance. If a close non-affiliate match is better than a weak affiliate match, the user experience should come first.

Brands, retailers, or affiliate networks cannot buy a guaranteed recommendation or ranking inside a personalized result. If Pick My Fit ever publishes sponsored content or paid placements, they should be labeled separately.

Limitations

Retailers control product descriptions, prices, sizes, colors, shipping, returns, and stock. Pick My Fit cannot guarantee that a product will remain available or fit correctly. Users should verify all product details on the retailer's website before buying.

Corrections

If a product link is broken, misleading, unavailable, unsafe, or a poor match for the generated outfit, contact ekvlabs@proton.me. We may update or remove links when issues are reported or discovered.