Lovi
Replacing Skincare Guesswork with Science-Based Evidence

After more than a decade building large-scale recommendation systems at Google and Yandex, Alexey Shagraev began questioning where that technological rigor mattered, seeking decisions with real-world consequences beyond optimizing engagement.

How did Alexey Shagraev’s experience lead to his interest in skincare?

Skincare emerged as the ideal proving ground. Most people choose skincare products by guessing—trusting labels, influencer recommendations, or brand narratives. But skincare decisions aren’t cosmetic trivia. They influence skin health, sensitivity, and long-term outcomes.

The problem is that the field is unstructured. Inputs are vague, outcomes are hard to measure, and even qualified experts often disagree. In a domain this fragmented, adding AI alone doesn’t solve the problem.

Before AI can advise on skincare, expert judgment must first be aligned into a shared, auditable structure. That insight is the foundation of Lovi.

Shagraev co-founded the company with Nadia Kapleva, Lovi’s medical director, who brings deep experience in cosmetic ingredient development. From the start, their goal wasn’t to build an AI skincare app. It was to turn skincare into a domain that could be evaluated consistently, by humans first, and only then by machines.

Together, they built an iOS platform that combines a cosmetics ingredient scanner, a face scanner trained by a medical board, and personalized skincare guidance delivered through a structured advisory system.
How does Lovi’s platform provide personalized skincare recommendations?

“Our app generates a science-based Lovi Score that evaluates products for an individual’s skin, suggests better matches when needed, and supports routines that adapt as users update profiles or track progress over time,” says Shagraev.

Importantly, all recommendations are brand-neutral and grounded in formulation logic and clinical reasoning, not marketing claims.

Underneath the product experience is Lovi’s core innovation: a rigorously defined methodology that aligns human expertise before any model is trained.

Why is aligning expert judgment crucial for Lovi’s system to work effectively?

Before Algorithms Scale Judgment, Experts Must Agree In areas like face analysis or formulation evaluation, two dermatologists can legitimately assess the same skin or product differently. Lovi addresses this by defining shared frameworks with clear criteria, scoring scales, and evaluation rules.

Members of Lovi’s medical board calibrate against the same methodology before labeling any data. Only then is that data used to train models. Outputs are continuously reviewed to ensure results remain clinically sound as the system evolves.

Rather than relying on brand claims, Lovi evaluates products across three core dimensions: effectiveness, skin-type suitability, and safety. This structure allows for nuance. A product can be effective but irritating, or gentle but limited in results. These dimensions form the basis of Lovi’s fit score, indicating how well a formulation matches a specific user.

Safety, in particular, is not treated as a generic checkbox. Lovi integrates regulatory standards across major global markets, ingredient review databases, and real-world market data. Ingredients with systemic risks, pregnancy-related concerns, allergen flags, or regulatory restrictions are handled conservatively. For sensitive skin, pregnancy, breastfeeding, or conditions such as dermatitis, recommendations automatically adapt based on formulation data and user context.

Trust Built Through Confirmation and Tracking Lovi’s face scanner is designed to do more than analyze. For many users, it confirms what they already sense about their skin by providing external validation that builds trust. It also enables consistent tracking over time by standardizing how images are captured, turning subjective impressions into measurable progress.

Behind the scenes, Lovi’s system blends supervised machine learning, expert-defined rules, and language models. Dermatologists, cosmetic chemists, and estheticians provide thousands of labeled examples describing how formulations interact with specific skin profiles. Algorithms are trained to reproduce that expert logic at scale and refined through real-world feedback on goals and outcomes.

The result is a system that scales professional judgment rather than amplifying marketing noise.

Lovi’s value isn’t personalization alone. It’s infrastructure. By converting an opinion-driven field into a measurable, auditable system, the company makes advanced technology genuinely useful in skincare.

That approach was recently recognized when Lovi received a Top AI Smart Skin Care App award, acknowledging not the novelty of AI claims, but the depth of methodological discipline behind the product.

What sets Lovi apart from other AI-powered skincare applications?

In a category crowded with “AI-powered” promises, Lovi stands apart by doing the unglamorous work first: building the structure AI needs to work.