How AI is reshaping experience design

For more than two decades, experience design has revolved around the interface. Designers mapped customer journeys, refined workflows and removed friction so people could complete tasks more quickly and easily. Whether someone was opening a banking app, shopping online or using an enterprise system, the experience effectively began when they arrived at the screen. That assumption no longer holds.

Today, much of the experience has already been shaped before anyone reaches an interface. AI has interpreted intent, summarized information and recommended actions before a customer visits a website or an employee opens an application. In many cases, the first interaction isn’t with the brand at all; it’s with an AI system acting on the brand’s behalf. Experience design hasn’t disappeared; it has moved upstream.

That shift is bigger than it first appears. It changes designers’ responsibility. Instead of focusing primarily on interfaces, we’re increasingly designing how AI provides recommendations, where human judgment belongs and what people need to feel confident acting on those recommendations.

Across the organizations we work with, one pattern keeps emerging: AI initiatives rarely struggle because the models aren’t capable enough. More often, the technology performs well while people continue doing exactly what they did before, checking every recommendation because they’re still accountable for the outcome. The issue isn’t intelligence; it’s confidence.

The experience begins before the interface

Traditional experience design assumed a predictable sequence. A person initiated an action, the system responded and the designer shaped the interaction between the two. AI changes that dynamic. People ask questions instead of navigating menus. They interrupt tasks, change direction and expect systems to adapt as the conversation evolves. By the time they reach an interface, AI may already have interpreted their request, considered available options and decided what information deserves to be surfaced. The interface increasingly becomes the place where a decision is revealed rather than where the experience begins. Most organizations are still investing as though the interface is where customers make decisions, but increasingly it isn’t.

Search engines answer questions directly instead of sending people to websites. AI assistants compare products before customers visit retailers. Enterprise copilots summarize documents before employees open them. Recommendation engines influence which products, services and content people see. The consequence is that organizations are no longer competing solely through better interfaces or smoother customer journeys. Increasingly, they’re competing to ensure their products, services and information are understood and accurately represented by AI systems.

Designing decisions, not interfaces

Human-centered design has always been built on a simple principle: Technology should support people rather than the other way around. That principle hasn’t changed. What has changed is the role technology now plays. Traditional systems waited for instructions. AI systems interpret context, generate recommendations and increasingly act on behalf of users. Design is no longer confined to the interaction itself. It extends into the decisions that shape the interaction before it ever happens.

This is where many organizations are still thinking about AI through the lens of traditional user experience. Much of what is described as AI experience design is still interface design with a chatbot attached. The interface has changed, but the underlying design approach hasn’t. AI produces an answer, while the interface simply delivers it. Designing for AI means making decisions that previously sat outside the design process. What information should shape a recommendation? How much uncertainty should people see? When should they intervene? What happens when the AI gets it wrong? These questions now have far more influence on the quality of an experience than the placement of a button or the layout of a dashboard. The interface still matters. It is simply no longer where the most important design decisions are made.

Human judgment becomes more valuable

One of the more interesting ideas to emerge from AI is the concept of the “reverse centaur.” The original centaur model came from competitive chess, where human-AI collaboration consistently saw better results than people or AI achieved alone. People defined the strategy while machines expanded what people could achieve. Enterprise AI is beginning to reverse that relationship. Increasingly, AI sets the pace while people review, approve and occasionally challenge its recommendations. In many workflows, humans have moved beyond directing every step and are instead validating decisions that have already been framed by the system. This shift is visible in customer service, software development, document review and operational decision-making. Teams are expected to review growing volumes of AI-generated work, often at a speed that leaves little room for thoughtful judgment. That’s where many organizations unintentionally undermine adoption. When people don’t have enough time or context to evaluate recommendations, they typically respond in one of two ways: they approve everything because the volume is overwhelming, or they verify everything because they don’t trust the system. Neither outcome delivers the productivity gains AI promised. 

The organizations making the greatest progress take a different approach. Rather than asking where humans should remain “in the loop,” they ask a more useful question: Where does human judgment create the greatest value? Those moments become deliberate design decisions. The experience slows down to match the “human tempo” of cognition where judgment matters and accelerates where it doesn’t. Designing for AI isn’t about removing people from the process, it’s about designing a better partnership between people and machines.

What leaders should do next 

The implications for leaders are less about technology than they are about priorities:

• First, stop treating AI as another digital channel. It changes how decisions are made, not simply how information is presented.
• Second, invest in the foundations before investing in more interfaces. Better information architecture, clearer business rules and well-structured knowledge will often deliver greater returns than another conversational front end.
• Third, redesign decision-making rather than simply automating workflows. Identify where AI should accelerate work, where it should pause and where human judgment materially improves the outcome.
• Finally, recognize that experience design has become a strategic capability.

For years, organizations differentiated themselves through better interfaces and more intuitive customer journeys. Increasingly, they’ll differentiate themselves by designing AI experiences that people understand, trust and willingly rely upon. That is a design challenge, not a technology challenge.