The Era of
Agent-Centric Design

AI Travel Agents

As a digital designer, I use various AI tools daily to make my workflow more efficient, analytical, and creative. During a two-week trip across the Far East this spring, I used virtual assistants like ChatGPT and DeepSeek for personal travel advice – for recommendations on sights, restaurants, and practical transport and ticket tips. Upon returning, I realized that we hadn’t visited a single website during the entire trip. Perhaps the Great Firewall and the Google Search ban played a role, but more likely, websites simply turned out to be unnecessary.

The virtual assistants provided surprisingly relevant information, particularly when the request was quite specific. For example, we were looking for non-touristic, authentic Buddhist temples primarily used by locals for worship and rituals. We also searched for creative neighborhoods where elaborately dressed TikTok influencers pose. Our hotel criteria were even more demanding. Virtual assistants handled such specific requests well and produced tailored output.

At the same time, articles emerged proclaiming the end of UI Design. As we increasingly outsource internet browsing to virtual assistants, websites are being visited less by humans and more by agents. Interestingly, these articles are written by the pioneers and godfathers of Human-Centered Design, including Jakob Nielsen and John Maeda. Perhaps it’s an effort to save their beloved field, or more likely because these visionaries sense how generative AI is fundamentally changing the World Wide Web.

Hello Agents. Goodbye UI Design. RIP Accessibility

Three bold statements that summarize the future of the web according to usability expert Jakob Nielsen. His article of the same name opens with a dramatic line: “Once, humans navigated the web manually. In the future, AI agents will act on our behalf, browsing, clicking, and deciding. This shift marks the end of traditional UI design and accessibility, ushering in a future where agents are the primary users of digital services.”

UI design and accessibility, therefore, become less relevant. AI agents will replace user interfaces, and disabled users will simply choose an AI agent that suits their needs. (1) This transition will take some time. Nielsen describes an interim phase where the simple autonomous agents we currently use evolve into fully autonomous ones. Until around 2030, websites will be visited by both human users and AI agents.

From Obstacle Course to Teleportation

Creative technologist John Maeda also bids farewell to UI design in his most recent Design in Tech Report, presented at the latest edition of SXSW. He refers to the traditional user interface as an “obstacle course”; a metaphor familiar to anyone who has ever booked a flight or hotel online. According to Maeda, in the near future, a simple prompt like “Reserve flights and book a hotel” will teleport us directly to the outcome. UX evolves into AX (Agent Experience), where UI is minimized in favor of direct AI execution. In his report, Maeda poses an intriguing question that, he predicts, will continue to haunt us:

Graphical User Interface

Before we answer that question, it’s worth putting these alarmist messages about the end of UI design into perspective. The graphical user interface (GUI) emerged in the early days of the internet to make user experiences more intuitive and user-friendly via icons, buttons, and menus. Navigating interfaces through text-based commands was a clunky endeavor.

While Nielsen and Maeda are enthusiastic about a future where the interface disappears entirely, we must remember that the GUI is crucial to a pleasant and efficient user experience. Consumers don’t always want to be teleported to the end result via a simple prompt like “Reserve flights and book a hotel.” Browsing visual information is part of the experience. People want to view and compare products from multiple angles – this even applies to more abstract products like flights, where one might want to choose a seat.

The UI Unfolds in Conversation

User interfaces won’t necessarily become less graphical or vanish altogether. Instead, the paradigm of the website changes with the rise of AI agents. Websites are no longer sequences of linked static templates. In the era of AI agents, they become dynamic: the user interface unfolds step-by-step through conversation.

This eliminates the “obstacle course” Maeda describes. AI agents simplify tasks like information seeking, shopping, and booking. Task time is reduced. AI agents provide an efficient and personalized user experience. Many traditional UX patterns like breadcrumbs, snackbars, and chips, which now clutter the interface, will become redundant or used more sparingly. The evolution from UX to AX doesn’t make the interface ephemeral but marks the end of the current paradigm: a linear chain of web pages resembling A4 sheets with scrollbars and buttons.

Dynamic Design

Even in the early days of the web, it became clear that websites were viewed on various devices. The mobile revolution led to responsive and adaptive web design, making design a dynamic concept for the first time. The interface adapted to the user’s context, mainly their device.

For decades, there has been talk about personalizing user interfaces. Until recently, this remained mostly a buzzword due to the lack of methodology and infrastructure (DEPT Trends 2024). But with the rise of AI and ML tools, personalization becomes a reality. This is essentially the next step in responsive and adaptive design. The interface builds itself in real time during interaction with the user, it is dynamic. The way we search, shop, and book becomes conversational.

Design in an AI-First Future

As Maeda suggests, and as I experienced during my vacation, users will increasingly bypass “carefully crafted” web interfaces by using AI agents. Now that ChatGPT has announced a partnership with the e-commerce platform Shopify that enables agents to purchase products on users’ behalf, this trend will likely accelerate. Conversely, Large Language Models like OpenAI and DeepSeek can be customized for brands using specific guidelines and training. These so-called Brand Language Models can be implemented on websites to improve the shopping experience, combining the power of LLMs with brand governance (DEPT Trends 2025).

The World Wide Web is becoming agentic. UX designers will design agentic systems and experiences. These systems live everywhere: as fully autonomous agents, virtual assistants on websites, and chatbots on social platforms. (2/3)

Designing the agentic experience happens not just at the surface (GUI), but increasingly at a deeper level, through orchestrating services and instructions. Information architecture, conversation design, operational thinking, and behavioral economics are key skills for the AI-first digital designer, who becomes something of an agent choreographer.

Design in the Age of the Algorithm

Since aesthetics is often underexposed in UX, I’ll close with a quote from American conceptual artist Sol LeWitt. He outsourced the creation of his artworks to a team of assistants, who, based on his instructions – what we might now call algorithms – executed the art. Creativity and craftsmanship are expressed in the idea and the code. (4)

Take the first steps into the Agentic Web

• Optimize your website by making it accessible to AI agents.
• Implement AI agents in your website or mobile app and make visitor interactions relevantly conversational. This will make interactions easier and more personal.
• Explore and prepare for new partnerships and integrations between LLMs and, for example, e-commerce platforms.
• Identify complex processes in customer interactions – the so-called bottlenecks – and simplify them through the use of AI agents. Time on task can thereby be reduced by an average of more than 50%.
• Automate your design workflow, for example for social media posts, by developing a generative brand tool. Train and equip AI agents with your brand-specific guidelines to ensure that generated assets are no longer fragmented but consistently on brand.
• Organize co-creation sessions between developers and designers to design the future user interface and product experience for collaborations between different agents and human users.

References

• Practicing CRAFT: Keeping Ahead of the Machine – John Maeda
  https://www.linkedin.com/pulse/practicing-craft-keeping-ahead-machine-john-maeda-dh33c
• Building Applications with AI Agents: Chapter 3 User Experience Design
• For Agentic Systems – Michael Albada – O’Reilly Media, Inc.
  https://learning.oreilly.com/library/view/building-applications-with/9781098176495/
• 2025 Design in Tech Report at SXSW – John Maeda
  https://johnmaeda.medium.com/autodesigners-on-autopilot-88c5b07609b9
• Hello AI Agents: Goodbye UI Design, RIP Accessibility – Jacob Nielsen
  https://www.uxtigers.com/post/ai-agents
• No More User Interface? – Jacob Nielsen
  https://www.uxtigers.com/post/no-more-ui
• The Evolution of AI Products – Luke Wroblewski
  https://www.lukew.com/ff/entry.asp?2096
• No AI agents were used in the writing of this article, except for occasional sentence structure checks.