Returning to Paris for the 10th anniversary of VivaTech felt like looking through a time capsule of innovation.
Perfect Corp. has participated in VivaTech throughout its history. During that time, I watched the event grow alongside many of the technologies presented there. Ideas that once attracted crowds because they seemed futuristic are now becoming part of the infrastructure through which people work, communicate, and shop.
This year’s gathering made one shift especially clear to me: The next phase of innovation will be defined by how intelligently different technologies work together.
That lesson is particularly visible in beauty technology.
BEAUTY TECH GREW UP
When virtual try-on first emerged, its ability to place lipstick, eye shadow, or other products on a person’s face in real time felt almost magical. It created an immersive experience and showed consumers something conventional e-commerce could not: how a product might actually look on them.
At that stage, beauty technology was often viewed as a spectacle. It attracted attention and increased engagement, but many companies still considered it an optional enhancement rather than a core part of the shopping journey.
The next stage was utility.
Virtual try-on became more realistic and commercially useful. AI skin analysis helped consumers better understand concerns such as dryness, spots, wrinkles, and pores. Shade-matching technology addressed one of beauty e-commerce’s most persistent problems: finding the right foundation without testing the product in person. Technology now helps the customer make a more informed decision.
Today, beauty technology is entering a third phase: infrastructure.
Capabilities that once operated as standalone tools are becoming connected across product discovery, consultation, personalization, visualization, and purchase. Consumers simply expect a brand to understand their needs, recommend something relevant, and give them confidence before they buy.
This represents an important leadership lesson: A technology category reaches maturity when consumers stop noticing the technology and start expecting the outcome.
FROM ASSISTED COMMERCE TO AGENTIC COMMERCE
For the past decade, beauty technology has assisted customers in navigating online stores. The customer still does most of the work: opening different pages, applying filters, launching individual tools, comparing products, and determining which recommendation to trust.
Agentic commerce begins from a different premise. Instead of asking consumers to understand the retailer’s catalog and digital architecture, an AI agent starts by understanding their intent.
A shopper might say, “I need a foundation for my skin tone that will work in humid weather” or “Help me create a skincare routine for dryness and dark spots.” Fulfilling that request may require several capabilities: interpreting natural language, analyzing an image, assessing skin attributes, matching products, explaining the recommendations, and allowing the shopper to visualize the result.
An agent can coordinate functions on the consumer’s behalf. That is the fundamental change from assisted to agentic commerce. Search asks consumers to understand the catalog. Agentic commerce asks the system to understand the consumer.
THE SMARTEST AGENT WILL NOT WORK ALONE
There is an assumption that the future will belong to one universal AI assistant capable of answering every question and completing every task. I believe the more likely future is an ecosystem of agents with different areas of expertise.
Beauty is an unusually complex category. Accurately assessing skin concerns, identifying undertones, matching foundation shades, recommending compatible products, and creating a complete look require specialized models, data, and domain knowledge.
A retailer’s general shopping agent may recognize that a customer needs foundation, but not possess the specialized intelligence to evaluate her skin tone and recommend the right shade. Instead, it could connect with a beauty agent that performs the analysis and returns the appropriate recommendations.
Smart AI agents need to know which expert to ask.
This agent-to-agent model could become the connective tissue of digital commerce. Brands and retailers could have their agents call upon trusted domain agents whenever a customer’s request requires deeper expertise.
In beauty commerce, this could mean a specialized agent interpreting a consumer’s request, consulting the appropriate diagnostic, matching, and visualization capabilities, and returning a coherent recommendation. Real innovation is the orchestration of multiple forms of expertise around a single customer goal.
That distinction matters. For years, companies have added digital features one at a time. Even when each tool performs well, the overall experience can still feel fragmented because the consumer must understand when and how to use each one. Agentic commerce reverses that burden. The technology organizes itself around the customer’s objective.
DESIGN THE NEXT DECADE
As business leaders evaluate agentic commerce, the starting point should be asking what decision the customer is trying to make, and what expertise is required to help her make it.
Companies must also resist the temptation to build systems that appear to know everything but lack depth in the moments that matter. In specialized categories, trust will depend on domain intelligence, transparency, and the recommendation quality.
Ten years ago, we came to VivaTech to demonstrate what beauty technology could do. Since then, beauty tech has progressed from spectacle to utility and from utility to infrastructure.
The next decade will be about connection: technologies communicating with one another, specialized agents combining their expertise, and commerce experiences adapting dynamically to individual intent.
The future will belong to the AI agent that can assemble the right intelligence at the right moment and make the complexity behind that collaboration feel effortless to the customer.
Alice Chang is CEO and founder of Perfect Corp.