Beyond Technology, Toward Connected Experience

When we talk about AI beauty tech, what typically comes to mind is individual capabilities: the accuracy of a skin-diagnosis model, the precision of a recommendation engine. The field has advanced considerably over the past several years, and with the broader acceleration of AI technology, expectations for beauty tech have risen accordingly. Image-based skin analysis, ingredient efficacy-driven recommendations, and conversational interfaces are all familiar territory by now. The technologies that quantitatively analyze skin condition, automatically detect issues, and suggest suitable products for each customer have already reached a meaningful level of maturity.

However, it is harder to say that the quality of experience customers actually feel has advanced at the same pace. Customers now have access to more varied and precise information about their skin, yet few have experienced that information translating into concrete action on their part. Even after receiving a diagnosis, the question “So what am I supposed to do now?” tends to go unanswered.

What our team has focused on, then, is not the performance of technology in isolation, but the way technology translates into experience. Today’s customers want more than just a result; they want to understand it, make choices based on it, act on those choices, and track the changes that follow. It is only when that sequence flows naturally that beauty tech becomes meaningful. That is precisely what Amorepacific’s AI beauty tech is working to build: a structure that holds this entire flow together.

The First Step in Customer Experience: Measurement

Image-based skin measurement is frequently cited as the starting point of beauty tech. A wide range of studies have demonstrated the use of deep learning models to assess skin conditions.

At Amorepacific, we have developed our own technology in this space: Dr.AMORE, which uses photos taken on a smartphone to automatically analyze skin conditions, including wrinkles, pigmentation, pores, and redness, and delivers a scored result. With no dedicated equipment required, it achieves accuracy exceeding 90% compared to professional skin measurement devices. Beyond the research paper and demo stage, this technology has already been deployed across multiple Amorepacific brands and service touchpoints, with real customers actively using it today.

The significance of mobile image-based measurement lies not only in its accuracy but in the foundation it provides for customers to recognize and manage change in their own skin over time. As repeated measurements become possible anytime, anywhere, customers begin to understand their skin’s transformation over time, and that understanding naturally flows into product choices and care methods. Measurement technology is shifting from being ‘a feature that delivers results’ to being ‘the touchpoint at which customer experience begins.’

Another Axis for Understanding the Customer: The Questionnaire

If image-based analysis belongs to the realm of the objective, the questionnaire is the means by which a customer’s subjective state and experience are captured. The questionnaire covers what images alone cannot reveal: moisture-oil balance, sensitivity, reactions to irritants, how skin changes in specific environments, everyday skin concerns, lifestyle, and dietary habits.

It is only when the physical state captured in images is combined with the subjective state perceived by the customer that a complete understanding of the skin becomes possible. The questionnaire matters not only as a data-collection tool but also as an opportunity for customers to reflect on their own condition. The act of answering questions is itself an experience, one that shapes subsequent product choices and care methods.

There is a growing effort to refine this questionnaire experience, whether by tailoring it to specific skin types, such as sensitive skin, or by structuring the questions around the customer’s primary concerns. One note of caution: a long and demanding questionnaire can reduce response accuracy. The key challenge at this stage is drawing maximum understanding from the minimum number of questions.

Where Measurement and Questionnaire Meet: Diagnosis

It is the combination of measurement and questionnaire that completes what we call ‘diagnosis.’ Diagnosis is the starting point from which all subsequent customer experience flows. The same measurement result can carry different meanings depending on a customer’s living environment, concerns, and sensitivity level; the same questionnaire responses can be interpreted more specifically when combined with objective data. In other words, diagnosis is not simply the step that tells customers their current condition; it is the foundation on which the trust of the entire downstream flow, from recommendation and explanation to action, depends. If the diagnosis is imprecise, customers will struggle to accept the results as truly their own, no matter how well the later stages are designed.

This ‘combination of measurement and questionnaire’ extends well beyond facial skin. The integrated approach is expanding into the many beauty categories customers encounter in daily life. For skin tone diagnosis, image-based measurement of color and evenness is combined with a questionnaire on everyday makeup concerns and preferred tones, enabling recommendations that go beyond simple color matching to align with how customers actually use products. In scalp and hair diagnosis, image-based measurements of sebum and scalp condition are paired with questionnaire responses on hair thickness and damage, yielding a fuller, more three-dimensional understanding of each customer’s condition.

From Report to Concrete Action: Recommendation

Recommendation is the area of beauty tech most directly tied to business outcomes. Customers do not stop at knowing their skin condition; they want to know which products are right for them. What they are looking for is less a list of products than a concrete answer to “so what should I do now?”: which order to apply things in, how many times a day is appropriate, and which combination suits their current condition. The key point at this stage is not how many products are recommended, but whether those recommendations are delivered in a form the customer can actually act on. A good recommendation, in short, helps customers decide on the right action for their own condition.

Recommendation is difficult to reduce to a simple matching problem. There is growing interest in approaches that link ingredient efficacy to specific skin conditions and concerns, as well as in knowledge graph-based recommendation systems that reflect relationships among products. These efforts should be understood as going beyond improving recommendation accuracy; they are attempts to provide the reasoning behind recommendations alongside the recommendations themselves.

Turning Recommendation into Conviction: Explanation

The moment a recommendation is presented, customers naturally begin asking: “Why this product?” “How does it compare to the alternatives?” “How exactly will it help my skin?” “Is there something that would pair well with it?” A conversational interface powered by a large language model (LLM) is a critical tool for answering these questions. The ability to explain analysis results in clear natural language and provide additional context in response to follow-up questions does more than add convenience: it builds trust. The kind of interaction in which the direction of explanation shifts with each new question creates a customer experience of an entirely different order from static information delivery.

Toward an Integrated Customer Experience

The individual elements covered so far — objective image-based measurement, subjective questionnaire, personalized recommendations based on diagnosis, and explanation of those recommendations — ultimately come together as a single continuous flow. The beauty tech industry is increasingly moving to integrate this flow into an agent architecture: a structure in which the system plans the necessary steps based on the customer’s information and responses, calls on the appropriate tools, and carries out actions in sequence.

The vision is a structure in which a customer uploads a photo, the system presents any necessary follow-up questions, recommends and explains suitable products and care methods based on the diagnosis, and extends through to future routines and ongoing management. Building that continuous flow is an entirely different problem from developing any individual technology within it. Each stage working well in isolation is one thing; having all of them connect as a single integrated experience is quite another. A successful architecture requires simultaneous consideration of real-time sensing and input data quality management, beauty profile utilization for personalization, recommendation accuracy and reliability across different skin conditions, explainability of reasoning, token costs, and service speed, all at once.

Customer behavior patterns are also shifting with the times. Rather than investing time in searching and comparing information as in the past, customers increasingly prefer experiences in which information is delivered to them in personalized form from the outset, signaling a shift in how customers make decisions. And as the flow of customer experience becomes more natural, customers stop noticing the technology. AI beauty tech becomes something they simply experience as part of daily life.

What Amorepacific’s AI Beauty Tech Is Building

From this perspective, what we are building is not a collection of individual technologies; it is a structure. How strong a model you have is now the baseline; the more important capability is how naturally you can connect a customer’s understanding, choices, and actions into a single continuous flow. Amorepacific’s AI beauty tech currently stands on par with the latest research at the global frontier. The accuracy of mobile-based skin measurement, the precision of combined image-and-questionnaire diagnosis, and the quality of LLM-driven explanatory recommendation are all in step with the most recent advances in the academic literature, and in some areas, it is fair to say we are a step ahead in terms of real-world service deployment.

‘Building a structure that connects customer decisions to action.’ I believe this is the phrase that best describes where beauty tech stands today. And this structure is not something that ever reaches a finished state; it is continually being refined and expanded through real customer experience.

The Next Step — Building It Together at Amorepacific

This flow will continue to take more concrete shape through close collaboration with business teams on the ground. However strong the underlying models and technical architecture may be, it is the people working directly at customer touchpoints who best understand the context, the moment, and the manner in which these capabilities should be brought to bear. When the field intuition and deep customer understanding accumulated across brands, commerce channels, CRM, and offline stores come together with technology, that is when beauty tech will be completed as a meaningful experience.

Our team is eager to collaborate with more departments and a wider range of brands. The conversation about which touchpoints hold which challenges, and which customer experiences you want to make more precise, is the very process through which technology gets shaped into something real. If you have new ideas, experimental initiatives, or even small problems to work through together, please feel free to reach out at any time.