Yoni Expert Case Study — Multimodal Vision AI Diagnostics & Sales

Online buyers were excluded from expert skincare advice, leading to low conversion rates, while generic chatbots failed to provide necessary skincare context.

• Customers could not get product recommendations without an in-person visit. Online buyers were completely excluded from personalised advice.
• Generic chatbots could not understand skincare context: fabric texture, skin plane, dryness patterns, pigmentation. Automated advice was useless.
• There was no way to connect a diagnosis to a product in stock. The customer had to bridge the gap manually, and most did not.

Built Nika Yoni, a custom-branded GPT-4o Vision agent that analyzes skin tone and texture from a photo, instantly matching the diagnosis to Shopify products with direct purchase links.

1. 1Built Nika Yoni, a custom-branded AI beauty agent deployed on expert.yoni1992.co.il with visual identity matching the Yoni Cosmetica brand.
2. 2Integrated OpenAI GPT-4o Vision to analyse user-uploaded photos. The agent reads skin tone, texture plane, dryness, pigmentation, and surface condition from a single image.
3. 3Implemented multi-layer context understanding: fabric structure analysis, surface geometry recognition, contamination mapping. Nika diagnoses the exact microenvironment, not just skin type.
4. 4Built a vector embedding layer that semantically matches each diagnosis to specific Yoni Cosmetica products. Product-precise, not generic category advice.
5. 5Embedded direct Shopify Storefront purchase links inside every recommendation. Customer gets personalised advice and clicks straight to checkout without leaving the conversation.