Patrick Neeman on uxGPT, Shifting UX Roles, and Business Impact
Patrick Neeman is Director of UX Design, AI Experiences at Workday. He is the author of uxGPT: Mastering AI Assistants for UX Designers and Product Managers.
Known for his approachable style, Patrick has carved out a niche in the world of user experience as a straightforward truth sayer through his insightful teachings, practical methodologies, and a knack for making complex concepts accessible and understandable.
With a career in high-profile roles working in mostly enterprise environments including financial spend, human capital management and contract intelligence, Patrick has led UX teams for over 15 years at Evisort, Knowable, Icertis, Apptio and Jobvite and in roles ranging from Director to Vice President. His ability to blend user research, interaction design, and business strategy has consistently resulted in increased engagement, and the occasional successful exit for the business.
uxGPT – Behind the Book
The X-Mentor: I first met Patrick during his time as Director of Product Design at Apptio, where he played a pivotal role in building the design team and leading the design of the company’s flagship product. From the beginning, it was clear that Patrick exemplified the rare “player-coach” mindset—equally effective as a hands-on practitioner and a strategic leader. His book, uxGPT, is a perfect embodiment of that approach: both a powerful tool for practitioners and a clear, thoughtful guide to the core principles of UX design in the age of AI.
Good morning, Patrick. Welcome to the X-Mentor.
Patrick: Nice to be here.
The X-Mentor: Let’s talk about your new book, uxGPT. You’ve made AI prompt writing more accessible and practical for UXers, which is something many in the UX field are just beginning to explore. Through your maker approach, you’ve brought others along as you’ve discovered how it really works. Today, we’ll be talking about key insights from the book and real-world applications that have been applied with your team.
But first, I’d like to ask what inspired you to write this book now?
Patrick: Well, the company I was working for at the time, we were playing with OpenAI on December 1st, 2022. We started playing with it, figuring out how we could apply it to the legal tech world. And then I realized early on when I started playing with it that we could use it for user experience activities. We started using it as a way of accelerating user experience because we didn't have a large design team and we really wanted to get the designers exposed. Later, I did a presentation in April 2023 on how to use generative AI for use experience activities and the models weren't nearly as good then. However, it was still good enough to where I could replicate a whole feature using GenAI.
I was showing people how to use ChatGPT for user experience and I basically walked about 75 people through, saying hey, this is what you could do, and it was fun and a true exploration. So, I was waiting for somebody to write the book. Years ago, I got offered a book deal by O'Reilly and I was just like I don't feel like writing a book now. But I kept waiting for somebody to write this book, and they weren't writing the book. And well, I said to myself, I better write this book.
So, in 2024, I sat down and wrote a book using ChatGPT, and I ended up finishing with Claude describing the process of how you would use AI assistance for user experience and product management activities. I avoided the prototyping part because the tools weren't ready for that yet. And that's how I just went about it and released it in October of 2024.
It was such an interesting experience because I felt like I had to live writing it using an AI assistant, to actually understand the process.
One of the interesting things about the methodologies in the book, it was literally my team at the time. We're using these methods to do user experience. We're using it for competitive analysis and we're using it for inspiration, and I wrote a billing specification for a feature in all of four days, and the first draft was used for engineering.
It really proved the fact that you’ve got this LLM and they have all this knowledge that's already out there. And let's be honest, a lot of this stuff with user experience is very predictable. There are areas that are more complex where we're delving into problems, like how you switch (i.e., orchestrate) AI agents. But for a lot of the work that I do in enterprise UX, it's usually tables and forms for many screens.
“I don't see a point in spending time and energy coming up with new patterns
that already exist with Jakob’s law as a north star.”
I don't see a point in spending time and energy coming up with new patterns that already exist with Jakob’s law as a north star. Many public contracts, such as those in the EDGAR SEC database, are included in large language models, providing a rich source of domain knowledge about legal agreements with helped our research. Since these contracts are publicly available and widely used to train models, designers can leverage them to better understand the nature of the work and the jobs to be done.
So, that was the impetus for the book. It was about showing this is something that can accelerate these user experience processes. And it's proven correct.
The X-Mentor: How did you develop and refine the prompts in the book? Was it trial and error, or did you codify a method using tools like ChatGPT to tell you what it could do? E.g., I’ve used the following prompt for ChatGPT: “tell me how to write a prompt for financial analysis.”
Patrick: I well, a lot of it was premise-based off a framework I developed while writing the book. We tried to come up with a standard framework about how we structure the prompts. One of the things I've learned about all these models is if you use consistent language, you get much better results. Because I've been writing for such a long time, is I tend to write in a very consistent voice and tone, and I use that for the prompting to get better results.
Everybody keeps talking about the hallucinations they get. And I just don't get them as much. I don't know why, other than I'm consistent about the way that I write. The other reason I wrote it that way was because I was trying to give everybody a scaffolding approach about learning the prompts.
The X-Mentor: In the process of refining prompts with your UX team, did any unexpected use cases emerge? Something you hadn’t anticipated the tool could do? Something that made you think, oh you can do this with ChatGPT? I had no idea!
Patrick: That's a really good question.
Most of what I’ve discovered about LLMs, like vector representations of words or their probabilistic nature, didn’t surprise me, since I’m already familiar with how they work. That said, I don’t think many people are aware of these underlying concepts.
One thing that surprised me about double shot prompting and scaffolding is how effective a layered, step-by-step approach can be, like a layer cake: start with one prompt, then go deeper with each layer.
I learned this structure from Carl Chatfield, a fantastic instructional designer we both worked with, and later reinforced it while teaching at General Assembly. It’s a powerful teaching method. And it works perfectly for guiding LLMs through prompt refinement and editing.
Whether using Lovable AI (an AI-powered app development platform) or working on other projects, I naturally apply that layer-cake approach. It’s become second nature.
The X-Mentor: Do you see a breakthrough on the horizon—like the GUI was for the web—that could trigger accelerated growth in AI beyond current prompt scaffolding? What kind of modern interfaces might drive that shift?
Patrick: I'm going to take this in a slightly different direction. I was talking with Scott Jensen recently, and he made an interesting point: many AI-generated experiences today are either average or slightly above average. Honestly, for much of what we do, basic CRUD operations—create, read, update, delete—still work really well. It's what people understand, and I don’t think we’ll ever fully move away from it. People have been trained in these patterns for 30-plus years, and for many applications, they just need something that works so they can complete their tasks.
This reminds me of something John Dvorak, the old PC World columnist, used to say: “the command line is the most efficient UI.” What we have now with LLMs is essentially a modern command line—just in natural language. In some ways, I think we’ve swung the pendulum too far in that direction. People still want more guided, intuitive experiences.
“What we have now with LLMs is essentially a modern command line.”
I believe we’ll start seeing more interfaces that anticipate the user’s next step; predictive, adaptive systems that try to guess what you’ll need next. Sometimes they’ll be right, sometimes wrong, and that’s okay. This new paradigm gives us space to experiment, and that’s where I think we’re headed.
The X-Mentor: Where do you see AI-generated output adding the most value to UX? And where should designers be especially cautious about over-relying on it?
Patrick: I think one of the key areas where caution is necessary is in how research results are returned. It's essential to always have a human in the loop to double-check the outputs. I believe it's a wonderful tool for generating new ideas and exploring the incredible range of possibilities that exist.
One way we're using it is through a custom GPT for a synthetic five-day design sprint. It walks through the five-day process popularized by Braden Kowitz and outlined in the Sprint book. It conducts interviews, explores aspirational designs, and generates user stories that can be fed into an AI prototyping tool. It includes the full back-and-forth interview format and more.
That said, while I support thinkers like Erika Hall, I would never use this to replace real user research. But would I use it to generate ideas or as a double-check? Absolutely. In fact, we recently ran a design sprint alongside a synthetic version, and the findings were fascinating, especially comparing how each approached the challenge.
The key takeaway is that we always double-check the AI-generated work to ensure the sources are reliable and accurate.
I look at myself as more of a curator and so there are a lot of places where we should be writing, but we must double-check what it says, right? And that's another area to watch for.
The X-Mentor: You mentioned the importance of staying grounded in “real user research.” Can you walk me through how you developed your prompt pyramid approach for user research? What specific challenges are they designed to solve for teams?
Patrick: I think the uxGPT book is both a snapshot of a moment in time and a preparation tool for the new world we're entering. Teams are becoming more compressed and more fluid, especially in terms of the talent stack. Roles are blending, and even the definition of "designing" is evolving.
Claire Vo had a fantastic video presentation titled Product Management is Dead, where she talks about this exact “Generalist Specialist” shift leading to an “AI-powered triple threat.” Today, anyone can create a prototype.
I've done this myself; generating user stories, a prototype, then working backward into the research. It’s a new dynamic where anyone can participate, and we’re increasingly working in smaller, more diverse teams.
The uxGPT book solves two key problems:
It provides a collection of prompts that teams can use in their day-to-day work.
It teaches UX fundamentals.
This is something I really want to emphasize. It explains why user journeys matter, when to use analogous inspiration, how and when to apply personas, and how to conduct competitive analysis.
You mentioned frameworks, and honestly, in many modern environments, those frameworks are either skipped entirely or poorly articulated. A big part of how I’m editing the book—and how I’ll approach future editions—is focused on bringing that foundational thinking back. AI or not, we should have foundational thinking about how we approach user experience.
“This stuff isn’t hard. Talk to users, group them, understand their use cases, then go design. It’s not hard, but it is essential.”
You’ve worked with me, so you know how often we skip over the basics: what UX actually is. The book reinforces that through the prompts and the educational content. This stuff isn’t hard. Talk to users, group them, understand their use cases, then go design. One of the main messages I’m trying to drive home is: it’s not hard, but it is essential.
The X-Mentor: I'm aligned with your point about the importance of validating research quality. My concern is, as tools like ChatGPT become more accessible, we're seeing non-experts, particularly Product Managers, relying on AI output without knowing when or how to question AI generated research results. From your perspective, what are the key challenges in upskilling these users to critically engage with AI-generated research?
Patrick: Well, I think one of the things we should learn how to do is question everything.
I read a lot of news. I read The New York Times, The Washington Post, and others. Even with those sources, I still cross-check to make sure what they’re saying holds up. I tend to be a bit of an apologist and in that way—I question everything we do.
Right now, we’re going down the path of orchestrating agents, doing a lot of research, and applying generative AI. And honestly, I’ll say it: we don’t know what the right solution is, because we’ve never had to design something like this before, so we have to test it to make sure we’re presenting the right solutions to the users.
So much of what we’re learning requires us to keep an open mind. There’s a huge volume of information out there and staying open to what’s actually real is critical.
AI is Redefining UX Roles
The X-Mentor: As we rethink UX roles in light of evolving team dynamics and emerging technologies, where do you see the UX discipline heading over the next three to five years—particularly as we move beyond current generative AI architectures into more advanced capabilities like JEPA (Joint Embedded Predictive Architectures) that can reason, plan, and understand World Models, such as physics?
Patrick: I tend to focus more on the 'meat and potatoes' problems. Right now, I'm working more on prototyping and the nuts and bolts of how we get product out the door. I haven’t yet dug into some of the deeper approaches around using research and analytics, though I’ve been hearing a lot of conversation about it. One thing to keep in mind with this whole transition is that we’re only about two and a half years in—just under three years since these technologies have really been out in the wild.
The X-Mentor: I know! It seems much longer than that.
Patrick: Yeah, it feels like it's been longer, to your point. We keep saying we needed a two- to three-year runway, but two years ago, who could have imagined where we'd be today? My running joke is that nine months from now, we might have flying cars—who knows?
I’ve been cautious about diving too deep too quickly because I want to stay grounded and maintain a strong foundation. One area that really excites me is AI prototyping. At places I've worked, we've used tools like Figma and Axure RP (a rapid prototyping tool) which I really like for storytelling. Many AI prototyping tools now replicate Axure RP surprisingly well and are great for conveying ideas.
At my current company, we’ve been using Lovable AI (an AI-Powered full-stack development platform), and it stops traffic. When I show a working prototype and mention it only took me eight hours over the weekend, people are floored. That’s where we need to get—to faster idea validation. While design thinking still has a place in high-risk, long-term efforts, a lot of what we do should follow a lean startup model.
I call it a 'design-first' mentality. You start with AI prompting to explore the core idea you’re tackling, which allows design and research to run in parallel.
I refer to this as a dual-track approach, something I’ve used often in my career. While traditional methods favor a more waterfall-style research process, I don’t fully agree with that. I believe there’s real power in designers and researchers collaborating simultaneously. Like the musician-producer dynamic that Alan Cooper’s agency once championed. I think that model will be incredibly powerful in the emerging landscape we're heading into.
The X-Mentor: So, do you think AI is redefining the core value of traditional UX professionals?
Patrick: No, we’re not redefining the core value, it’s always been there. Much of our field has simply forgotten it. One thing I often tell people is that I'm essentially a product manager who thinks deeply about design.
But I've always focused on key questions like: How are we spending our money? How are we creating business value? Throughout my career, I’ve encountered organizations that overemphasized design craft at the expense of business outcomes.
At places like Apptio, we had to make tough, sometimes unpopular design decisions that prioritized the needs of the business and the user—even if that meant visible seams in the experience. And those seams were appropriate for our users.
“The core value we should always be championing is this:
How do we run our world like a business?”
Sure, I could have pounded the table insisting everything be pixel-perfect and consistent. But it wouldn’t have been worth my time, or the company’s money.
The core value we should always be championing is this: How do we run our world like a business?
The X-Mentor: Or as I say, Manage Experiences like a Business. Tap The X-Mentor to learn more. (Shameless plug!)
Patrick, many in UX have spent decades operating tactically—pushing pixels, building interfaces or design systems—often branding themselves that way. But to stay relevant, we need to shift toward strategic influence. Have you thought about what that transition looks like for UX professionals trying to move from execution to real strategic impact?
Patrick: Well, I'm already there.
The X-Mentor: Indeed, you are! So can you share with our readers, what’s the most effective path for UX professionals to gain strategic influence across their organizations?
Patrick: Understand the business. One of the things I often share are links that illustrate different points I’m making. I was in the enviable position of having more domain knowledge about the field I was in, with real-world context and paying customers, than almost anyone else at the startup—except for a couple of people. You wouldn’t believe how powerful that is. It allowed me to think and act like a product manager, speaking at length about users, customers, and business context—rather than just where to place a pixel.
Many designers never get to that level of understanding.
That’s what’s so ironic about generative AI: it offers a world of knowledge on any topic you want, and you can ask it questions. If you want to learn about the domain you’re designing for, all you must do is ask. That’s why I urge designers to deeply learn the domains they work in—whether it’s finance, consumer loans, legal tech, or marketing. AI can provide a strong foundation, E.g., how the industry talks, the business structure, the user journey. You can then validate this with SMEs inside your company.
Sure, AI will hallucinate sometimes, but so do people.
It’s still an excellent starting point. One of my colleagues, Julie Booth—who wrote one of my forewords—is often called the ‘Noam Chomsky’ of her organization. She’ll get a kick out of reading this interview.
“English language is now a programming language.”
The other thing that's going to be necessary for all these designers and PMs—and really for everyone—is having a strong command of the English language. English is now a programming language. In fact, any language is now a programming language, and you need to have a strong command of it. Many people don’t. The more data an AI model has about you, the more accurately it can replicate your voice and writing style—down to the point where it feels unmistakably personal.
The X-Mentor: You mentioned 'design thinking' earlier. In some organizations, I’m reflecting on my experience at IBM here, it has become almost a dogmatic religion. How do you see AI reshaping design thinking processes, and more importantly, how can we leverage it to accelerate outcomes and deliver value faster?
Patrick: I don't think the process itself changes entirely, but the way we approach it will significantly. One methodology I strongly advocate for is The Lean Startup approach, from Eric Ries’s book. It emphasizes 'Build, Measure, Learn.' It’s not about perfect solutions, but about putting out something at 80% and iterating. The key is to avoid irreversible decisions and to learn quickly from both internal and external users.
That’s how you get meaningful feedback and make real progress. Too often in the past, we’ve engaged in what I call 'UX theater'—a lot of activity with little business impact. That’s disappointing because it reflects a lack of business understanding. With today’s tools, we have a real opportunity to deliver value faster and show we’re deeply informed about the people we're designing for.
The X-Mentor: When Don Norman coined the phrase 'user experience,' it launched an entire field. Over time, new roles emerged that hadn’t existed before. As AI becomes more embedded in design, do you see any new roles on the horizon? What positions do you think will become essential as this shift accelerates?
Patrick: Whenever I look to the future, I always look to the past. One role I think is especially important, and likely to become even more valuable, is that of the Design Technologist. I had the opportunity to work with the very first design technologist at Amazon. He’s this amazing blend of research expertise, just enough design knowledge to be dangerous, and strong coding skills.
That role from the past is proving incredibly relevant again. On my team, he’s been delivering great solutions despite limited context—thriving in ambiguity, which is key. I'm starting to see similar roles emerge: one acquaintance just became an AI UX Engineer at Microsoft, focused entirely on prototyping. Another role I came across was titled Product Engineer, responsible for using AI to generate and validate UI components.
We're entering a new era where AI-driven prototyping is taking hold, and it's opening an entirely new class of jobs. Some designers will adapt and thrive in these roles. Others may struggle to keep up.
The X-Mentor: I want to quickly close the loop on that last point. How should design education, whether through bootcamps or university programs, evolve to prepare talent for this new AI era?
Patrick: Well, when it comes to business-ready designers, there aren’t that many. I taught my General Assembly class with that focus — more about design thinking and how to connect design with business goals. I’m not sure what everyone else is doing, but many of those students are still in the field, so it must have worked.
The truth is that a lot of design education has never really aligned well with business needs. That’s something schools need to rethink. They must redevelop their curriculum to bridge that gap, especially because design roles are becoming much broader in many organizations.
Shift from Craft to Business Impact
The X-Mentor: Design has traditionally focused on craft, but businesses expect measurable returns on their investment. How should designers reframe their value proposition to clearly demonstrate business impact?
Patrick: It's really for them to reframe it for the business because they have got to define it. One of the things everybody talks about is having a seat at the table. I've been lucky enough to have that seat and I’m telling you it's a food fight.
[Laughs!]
The X-Mentor: Designers who want a seat at the table have no idea what they’re asking for.
Patrick: Yeah, it's… I mean, I can sit at a different table now, one that doesn’t have John Belushi sitting to my right. But part of the challenge is that we must define the value of design for others. We need to say, "This is the impact we’re having," and the best way to show it is through real results: we designed the solution, and here’s the improved conversion rate, the increased engagement, or the growth in subscriptions.
Designers have never done a particularly good job of demonstrating that impact.
I'll even be a little spicy with this comment: it hasn’t really paid for designers to step up in many cases, so I understand the risks. Even when they’ve taken a backseat role in many organizations, the compensation has still been pretty good. So, many designers haven’t wanted to risk taking a more visible seat at the table—because doing so might put their paycheck at risk. But now, that complacency is going to affect their paycheck.
If we don’t have a seat at the table, we’re not influencing the work, and eventually, we’re not justifying our roles.
I was just talking about this with a friend, David Leahy, who’s on the CX side. Customer success earned their seat at the table by taking ownership of renewals and reducing churn. That was their deal with the devil, so to speak—it was clear, measurable, and tied to business outcomes.
What are we willing to own in order to earn—and keep—a seat at the table?
We need to figure out: what's our version of that deal? What are we willing to own in order to earn—and keep—a seat at the table?
The X-Mentor: What’s your advice to designers on quantifying their impact in a way that goes beyond arbitrary benchmarks, especially when traditional ROI doesn’t always resonate in design?
Patrick: Here’s the advice I like to give designers, especially when they’re putting together a portfolio piece:
Ask yourself, how did this make the world a better place? More specifically, how did it drive real outcomes? Did it lead to more user sign-ups, increased engagement, or better retention? Can you tell the story in a deep and meaningful way? There needs to be a clear link between what you did and the value it created; whether that's increased revenue, improved customer satisfaction, or achieving a key business goal.
For many designers, that’s hard to quantify. A lot of organizations don’t measure things that way, or the data simply isn’t accessible. But it’s getting better, certainly better than the days when we had to build our own usage tracking systems just to get basic metrics.
Still, that’s where designers need to go. In every conversation, the focus should be:
How does this improve the business?
The X-Mentor: Given that researchers already measure customer experience in detailed, tactical ways, how can research play a more catalytic role in demonstrating business impact—specifically by tying customer value directly to business outcomes?
Patrick: I actually think designers should do this. One of the things I always recommend is: if you want to truly understand your users and customers, go straight to Customer Success. Build a strong relationship there so you can get access to the numbers and establish real value.
Rachel McBrearty, who was the Chief Customer Officer at Evisort and is now part of the same ecosystem I work in, is a great example. She’s a former designer herself, which made working with her incredibly effective. I didn’t need typical one-on-ones with her because we were 100% aligned from day one. I could get access to users and customers instantly, whenever I needed it. And we always aligned on what to measure.
So, for designers who want to understand the business and create impact, the best place to start is with Customer Success—just like product managers do—to truly understand customer pain points.
The X-Mentor: We often talk about breaking down silos, and we’re seeing UX and CX increasingly converge into one continuous experience, especially around the shared context of the customer journey. How can journey-based thinking help unify these functions and drive end-to-end impact, from user experience to customer success?
Patrick: Yeah, and it enables what I call continuous learning. One of the things I struggle with when it comes to design thinking is that people often treat it like a waterfall process. You go do research, and that gives you context, and then you move on. But when you're working inside an organization, you should be learning all the time.
I made a comment just the other day that, at some point, I stopped checking usage analytics daily, because I already knew exactly what I would see. I had internalized the patterns. That’s what continuous learning looks like.
A lot of the designers I’ve worked with have reached that point too. When we worked together, that was our shared approach.
Research was never a one-time activity, it was ongoing. And to make that work, you must carve out time regularly to keep learning and doing the research, continuously.
The X-Mentor: Right. Continuously learn.
Patrick: Yeah, continuously.
The X-Mentor: "How can business leaders shift from prioritizing shareholder value to putting customer value first—using it as the primary driver of business success?
Patrick: The way we can do this is by first clearly defining the problem we're solving. From there, we can use tools like GPTs to help generate measurable outcomes. It's not rocket science. You simply say, "Based on this problem, we believe this is how the world will be a better place."
A lot of companies have done this well for a long time. They talk about things like lifetime customer value and repeat engagement. Many brick-and-mortar companies excel at this. One of the best conversations I’ve had on the topic involved Costco.
Thanks to their membership model, Costco has more customer data than just about anyone. That enables them to apply design thinking across the board. Their store managers are empowered to make decisions based on their local environments and customer context, which is amazing.
They don’t have a lot of designers, but they have a lot of people who think like designers.
Their stores are intentionally designed to drive sales and to create intrinsic value through product discovery.
In fact, many brick-and-mortar companies are further along in this kind of design-led thinking than digital companies. We've dropped the ball. Just walk into a supermarket and you’ll see slotting, optimized aisle layouts, and detailed strategies based on massive amounts of data.
Supermarkets operate on thin margins, yet they’ve managed to make the business highly profitable—if they do it right. And I guarantee most technologists never walk into a store thinking about its information architecture. Maybe they should.
The X-Mentor: Patrick, before we wrap up, is there a question you wish I had asked—something you were hoping to answer?
Patrick: I think the question I wish you would have asked is, do these AI assistants destroy the value of design?
The X-Mentor: Patrick, do these AI assistants destroy the value of design?
Patrick: No
AI systems can't reason—at least not in the way we understand human reasoning. That’s why the notion of having appropriate taste remains incredibly important in this new world. It's a core value we bring, and it’s not going away anytime soon.
The X-Mentor: Exactly. Yann LeCun and his team are working on JEPA (Joint Embedded Predictive Architecture), which aims to tackle planning and reasoning through world models like physics. The wildest breakthroughs are likely to be still ahead. So, as you might say, AI doesn’t have the ability to reason—yet.
Patrick, this has been such a pleasure to reconnect with you and learn more about your new book, uxGPT and the behind-the-scenes thoughts and process that went into its development.
Thank you for sharing with us today on The X-Mentor!
ABOUT THE AUTHOR(S)
Patrick Neeman is Director of UX Design, AI Experiences at Workday. He is the author of uxGPT: Mastering AI Assistants for UX Designers and Product Managers.
Greg Parrott is The X-Mentor and publisher of The X-Mentor on Substack.