The AI Economy: Redefining UX in the Agent Era

Welcome to "The AI Economy," a weekly newsletter by Ken Yeung on how AI is influencing business, work, society, and technology. Subscribe now to stay ahead with expert insights and curated updates—delivered straight to your inbox.

IN THIS ISSUE: Unpack all the insights from John Maeda’s Design in Tech 2025 Report, where AI isn’t replacing designers—it’s redefining the entire discipline. From the rise of agents to the shift from UX to AX (Agent Experience), Maeda maps out how design must evolve to meet an AI-augmented world.

The Prompt

At this year’s South by Southwest conference, Microsoft’s Vice President of Engineering and its Head of Computational Design and AI Platform, John Maeda, took the stage to unveil his 2025 Design in Tech Report. Now in its 11th edition, this report offers a rallying cry to designers, technologists, and business leaders alike: AI isn’t replacing designers. It’s reshaping design itself.

Drawing on more than a year of bookmarks and research and an assist from OpenAI’s GPT-4, Maeda outlined a vision for how generative AI and agentic computing are overhauling everything from UX to programming to professional creativity.

Here’s a look at seven key takeaways from this year’s report:

AI Is Not Replacing Designers—It’s Transforming Design Itself

Maeda doesn’t say human designers will be out of work thanks to AI. Instead, he contends the technology will continue to change how they work. Anyone who has been tracking the Design in Tech reports will recognize that he has been making the case for design’s evolution in the face of innovation. Early on, he pushed for designers to shift from focusing on classical design to computational design. And now, in a world populated by large language models (LLMs), creative professionals must adapt better.

In a world increasingly powered by AI agents, designers will find their roles redefined. While bots handle automation and execution, it’s up to human designers to shape the overall AI experience with creativity, context, and intent. At the same time, the creative process is being disrupted by AI tools, prompting a shift in how and what designers create. These professionals can leverage the technology to analyze large amounts of data quickly, generate initial concepts, prototype faster, or even explore multiple design variations rapidly.

AI Experimentation Is Getting Cheaper, Faster, and More Accessible

With token costs decreasing, there are fewer barriers for people to experiment with AI. Designers and developers can take advantage of this by building, testing, and iterating with advanced models in minutes‚ at a fraction of the cost.

“It doesn’t cost a lot to experiment with AI. It’s relatively cheap,” Maeda said. “For instance, last year, you’d pay $30 for GPT-4 and…$2.50 for GPT-4o…not bad, huh?” He highlights that reasoning, when introduced by OpenAI’s o1 model, was expensive. However, DeepSeek’s r1 model shows that it can be done cheaper.

Moving AI Away From Models to Agents

By now, it’s clear that autonomous intelligent agents are rapidly multiplying across the internet. Every day, you read about new agents being announced, such as from Salesforce and Microsoft. It’s proliferated so much that Maeda suggests it’s time for people to “stop saying models and just say agents.”

He sees loops—repeating processes in code—as being a critical mechanism in this area. Specifically, he believes that continuous loops are what help AI agents operate persistently and intelligently over time. And it’s because token costs have become more affordable.

From UX to AX

To Maeda, the concept of user experience (UX) has evolved for the AI age. Designers need to shift their focus towards something he calls the agent experience (AX). He describes it as something markedly different from UX in that it’s not about creating for users or people, but rather the AI agents.

“It’s not adding agents to your UX. It’s about the UX for AI. It’s to make it easier for AI to use software. Think about APIs, how the obstacle course of GUI sits on top of APIs. And so if AIs can talk directly to AIs instead of using a graphical interface, it’d be much easier.” He cites the LLMs.txt movement in which LLM-friendly instructions are attached to webpages to act as guides for the bots.

AI UX Best Practices Improve Trust and Usability

As AI systems take on more responsibility, trust and transparency become critical. This creates an enormous opportunity for designers in the field of AI UX. Maeda referenced his LinkedIn Learning course, “UX for AI Design Practices for AI Developers,” in which he has five important things to know: Post AI notices (disclose when you’re using the technology), suggest next steps to help with indecisions, display citations when AI is used, provide a status update when latency is an issue, and be receptive to feedback.

He calls out OpenAI’s o1 reasoning model as a good example of AI UX, saying it demonstrates how the model thinks and reasons. It’s a pattern that Maeda asserts “is here to stay.”

There are four distinct AI UX spaces that illustrate how we interact with intelligent systems. The first is chat, the familiar back-and-forth messaging format that has defined most AI conversations so far. The second is documents, where AI turns simple prompts into rich, “turbocharged” content—expanding ideas into fully-formed text with structure and depth. The third is tables, or AI-enhanced spreadsheets, where language models power dynamic data manipulation and logic. And finally, there’s the canvas — a freeform, two-dimensional space that allows for infinite, flexible exploration, enabling users (and agents) to work visually, spatially, and interactively.

Responsible AI Governance

The increasing capabilities of AI bring forth the need for responsible governance, a message many pundits, analysts, and AI critics have uttered over the years. Maeda warns that as agents are given greater access to tools, APIs, and system-level commands, AI must not be left unsupervised lest it potentially lead to accidental and malicious harm.

In his presentation, he pointed out that, given tools, bots can become increasingly powerful, execute tasks, and act independently. However, it can be challenging to monitor and produce errors that could affect systems.

He would stress that tools like evaluation frameworks are essential to prevent these outcomes. Maeda urges designers and developers to test AI not just for accuracy, but for safety, behavior consistency, and user impact. “You don’t want to hurt your users, and you don’t want to hurt your brand,” he declared.

The Importance of Human Adaptability

The final takeaway involves adaptability, the willingness to learn, unlearn, and relearn as tools evolve. Maeda urges designers to embrace change: “You have got to be adaptable as a human, because AI will augment the future, and we have to adapt while this happens.” The ways of doing design in the past may no longer be applicable in this AI era, so humans need to accommodate this new technological landscape. Maeda doesn’t position this as a threat to designers’ livelihood, but more like an opportunity for growth.

Maeda’s 2025 Design in Tech Report is more than a trend roundup — it’s a forward-looking vision of a world where AI holds greater agency, human creativity remains irreplaceable, and adaptability becomes the key to thriving in a rapidly evolving design landscape.

▶️ Watch the complete unabridged Design in Tech presentation from SXSW

Today’s Visual Snapshot

An infographic showing generative AI use cases between 2024 and 2025. Image credit: Harvard Business Review/data from Filtered.com

A look at the top ten generative AI use cases, according to Filtered.com. Between 2024 and 2025, people started to use AI more for personal development (e.g., therapy and companionship, organizing their lives, finding purpose, living healthier lives, generating ideas, and creativity) versus so-called “technical assistance and troubleshooting” purposes (e.g., personalized learning, exploring topics of interest, etc.).

Quote This

“I hope Copilot is a good CEO!”

— Microsoft co-founder Bill Gates joked at the company’s 50th anniversary event, where he was asked what he hopes Microsoft will have accomplished by its 100th. He would later add he hoped the firm would “lead the AI age [and] shape it for the good of everyone we work with.”

This Week’s AI News

⚙️ Hardware, Robotics, and Autonomous Systems

🔬 Science and Breakthroughs

Google created a new AI model for talking to dolphins (Ars Technica)
Why AI is better than doctors at the most human part of medicine (Bloomberg)

💼 Business, Marketing, Media, and Consumer Applications

Study: Immigrant founders are the norm in key U.S. AI firms (Axios)
I tested the AI that calls your elderly parents if you can’t be bothered (404 Media)
Netflix co-CEO: AI can make movies “10% better” and talent is using it to “do set references, pre-vis, VFX sequence prep, shot planning, all kinds of things..to make the process better.” (Deadline)
The real AI challenge for WPP isn’t scale, it’s control (Digiday)
Deezer says 18 percent of music uploaded to its service every day is AI-generated (Engadget)
Job seekers turn to AI tools to gain a competitive edge. It can also backfire (Los Angeles Times)

💥 Disruption, Misinformation, and Risks

Wikipedia is giving AI developers its data to fend off bot scrapers (The Verge)
The latest viral ChatGPT trend is doing “reverse location search” from photos (TechCrunch)
AI is making online shopping scams harder to spot (CBS News)
Venezuelan migrants relied on clickwork to survive. Now AI is replacing them (Rest of World)

🔎 Opinions, Analysis, and Editorials

The end of business as usual: How AI-native companies win (Information Week)
Vibing dangerously: The hidden risks of AI-generated code (The New Stack)
Swapping LLMs isn’t plug-and-play: Inside the hidden cost of model migration (VentureBeat)
AI startups, take note: How to patent your tech in the age of Alice Corp. v. CLS Bank International (Crunchbase News)

Featured Image: John Maeda's 2025 Design in Tech Report reveals how designers are adapting to the agentic era, creating experiences not for screens and humans but for bots. Image credit: Ken Yeung/Adobe Firefly

End Output

Thanks for reading. Be sure to subscribe so you don’t miss any future issues of this newsletter.

Did you miss any AI articles this week? Fret not; I’m curating the big stories in my Flipboard Magazine, “The AI Economy.”

Follow my Flipboard Magazine for all the latest AI news I curate for “The AI Economy” newsletter.

Connect with me on LinkedIn and check out my blog to read more insights and thoughts on business and technology.

Do you have a story you think would be a great fit for “The AI Economy”? Awesome! Shoot me a message – I’m all ears!

Until next time, stay curious!

Subscribe to “The AI Economy”

Exploring AI’s impact on business, work, society, and technology.

Beyond UX: Why the Future of Tech Is About Creating Interfaces for AI, Not Humans

The Prompt

AI Is Not Replacing Designers—It’s Transforming Design Itself