Paul Ford: Writer, developer & "fun Cassandra" on why everything is changing (but not how you think)

0:00Sometimes you can replace a designer with a robot that just says we need to do more research before the work can happen, right? And that's the thing I actually like about designers.

0:08Paul Ford:They're like, hold on a minute. We need to bring it in. We need to understand. And it can drive you crazy when you're pitching a project and trying to get it over the line. But they're right. We have to get context before this product can get moving. And it's going to start with mood boards and ideation, but it's going to end with us sharing a language with the customer. It's visual and it's words and it's texture and it's vibes and it's all of that. Paul Ford likes to call himself a fun Cassandra, someone who, like the priestess in Greek mythology, sees trouble coming, but unlike her, tries to make the warning as entertaining as possible.

0:41He's the writer, developer, and co-founder of the tech agency Aboard who saw Claude Code drop last November and immediately understood it was going to change everything, while finding, to his surprise, that most people around him simply weren't seeing it that way. That same instinct is what drew him into AI early. Where others hedged, Paul dove right in, vibe coding nonstop, running full enterprise subscriptions for his entire team, and building in earnest. But he's not a fanboy. He's a critical optimist who believes something important is happening right now while holding equal concerns about the companies pushing it, the students expected to learn from it, and the decades of hard-won knowledge that might be quietly evaporating in the rush.

1:25Paul is also an English major who sold an agency, a developer who thinks in prose, and he's a father of 14-year-old twins. He is one of the most multidisciplinary thinkers that we've encountered on the show. He won the National Magazine Award for writing an entire issue of Bloomberg Businessweek dedicated to explaining programming to mass audiences. A 38 ,000-word essay called What is Code? a regular contributor to Wired magazine and published in the New Yorker and MIT Technology Review. He's one of the rare writers who can make the inner workings of software feel urgent and human. So he's exactly the kind of thinker this moment needs.

2:07Someone who can write code and read the room and who cares about quality as much as velocity. He can also make you laugh while explaining why you should probably be just a little bit worried. This is Design Better where we explore creativity at the intersection of design and technology. I'm Eli Woolard. And I'm Aaron Walter. If you're hearing this, you're not currently on our premium subscriber feed. Design Better Premium subscribers enjoy weekly episodes. That's four episodes per month rather than just two. And all of them are ad free. Plus, you'll get an invitation to our monthly AMAs with the smartest folks in design and tech.

2:43And if you subscribe at the annual level, you'll also get our toolkit, a collection of our favorite design and productivity tools like Perplexity, Miro, Read AI, and more. You'll hear a preview of this episode, but if you'd like to hear the full conversation, please consider becoming a premium subscriber at designbetterpodcast.com slash subscribe. The podcast is available to everyone through our scholarship program. So if you can't afford a subscription, just shoot us an email at subscriptions at thecuriositydepartment.com. We'll help you out.

3:25We'll return to the conversation after this quick break.

3:32So good, so good, so good. Everything you want for summer is at Nordstrom Rack stores now and up to 60 % off. Stock up and save on the brands you love like Vince, Sam Edelman, Frame, and Free People. Join the Nordiclub to unlock exclusive discounts, shop new arrivals first, and more. Plus, buy online and pick up at your favorite rack store for free. Great brands, great prices. That's why you rack.

4:02And now, back to the show.

4:09Paul Ford, welcome to Design Better. Thank you so much for having me. We are so excited to have you. Our mutual friend, Meredith, at the New York Times, invited you to their wonderful location. And we heard rave reviews. So we thought we've got to get you on the show. And I have been listening to some interviews, one most recently with Slate Magazine, I think it was, that you're on. And there was something that caught my attention where you called yourself a fun Cassandra. And I thought that was great. And for folks that don't know the Greek mythology reference, Cassandra was this figure that made these dire predictions in not such a fun way, but you spin it in a fun way.

4:47So maybe you could speak to that a little bit and how that shapes your attitude towards our current era of AI.

4:55Paul Ford:Yes. Okay, so I make predictions sometimes, and sometimes they're pretty good. Like I can go back and find, you know, like I saw that this was going to be a big deal, and it'll be like some boring web thing. I'm wrong a lot too, by the way. I'm like, I can absolutely find tons of places where I made very bad predictions. But every now and then I kind of nail it. And AI hit and Cloud Code hit and it hit around November. And my background is as a developer and a writer. And I was just like, oh, this is going to be a really big deal. And I was really surprised, I think, to find that people weren't seeing it in the same way that I was.

5:29Paul Ford:And so I felt that I'd fallen back into fund Cassandra mode. The worst prediction I ever made that was also the truest was I saw Gamergate happening. And it was all this conflict online about just if you don't know what it is, go read the Wikipedia page. And I just went like, if you want to know what the future of American electoral politics looks like, it'll be like Gamergate for the rest of your life. And it was like the saddest thing to be right about. I like to pick a culture. I'm a very anxious person. And so I poke around. So I think with AI, I feel very much like the fun Cassandra because I'm like, I think this is a really interesting, empowering tool.

6:02Paul Ford:I think we should be very critical of it. I don't know if I love the companies that are pushing it as hard as they are. I think the entire tech industry, Eli, you're in Stanford area, you've seen how ridiculous the billboards are. But at the same time, there clearly is something here. Work is getting done very quickly and things are changing. And so it's not even a prediction. It's an observation, but it still didn't register as good news for a lot of people. Before we hit record here, you said some stuff that very much resonated with me. A little setup here for our listeners. You run an agency called Aboard.

6:39Co-run. Co-run. All right, so you're a key figure in this business. And it's a rather forward-thinking business in that it's an agency, and you're using AI to provide services and build software for your customers. Now, most folks would think like, hey, that sounds like a bad business model because AI isn't putting us out of a job and you're sort of embracing that it is not. It's actually accelerating. We can come back to that topic. So set that to the side. But you said that with your clients, the way that you're using AI is making the ideation, like let's explore the ideas, accelerate like crazy fast.

7:22And then the last part of actually building the thing out, it's as slow as ever, in fact, maybe slower. I'm experiencing this right now. I'm working on a small project. I'm building my own website. And once you kind of get through the clear ideation, the building is still just a slog. It's just still a lot of shit you got to do. I'm curious to hear how this is playing out for you and your colleagues professionally.

7:47Paul Ford:It's really tricky and it's really hard to articulate. And I think it's because people are seeing these wild results at the prompt. You're typing things in and you're getting a PowerPoint deck or it's writing you all this code. And it's incredibly fast. It's incredibly high velocity. And it's very easy to get a very standard thing, even a pretty big standard thing to exist. So what's a standard thing? A standard thing is an outline for a research report, a business case for something, a summary of a lot of documents, a PowerPoint that looks kind of like all the other PowerPoints. And a lot of times that's actually better than many of the materials that you might have had before.

8:22Paul Ford:And so that gets really exciting and motivating to people. But I've been kind of vibe coding nonstop since November, less because I need to for work and more because I'm sort of on the hook to understand. And what I'm finding, we're finding this throughout the company because I'm not the only one. There's lots of people building. We sort of went in all full enterprise subscriptions. Every engineer in the firm, everybody we work with is using this stuff, even though a lot of the things we're delivering are not super AI, but are just classic software, just faster. And we had a client come in in the sciences, and they built this very intense machine learning model.

8:54Paul Ford:And they're having trouble getting people using it. And they're like, could we just get an LLM in front of this? And could we do some other stuff? One of our heads of engineering built out a really nice workflow-driven, multi-step, wizard-style interface for this. It's just very mature. He's been at it forever. And it feels dangerous to show it to them because in three days, I can get you kind of a miracle. But after that, it just becomes product again. We got to have meetings to talk about what we want and we got to iterate. The coding is really fast, but the thinking is actually not that much faster.

9:28Paul Ford:It's maybe a little bit faster or possibly a little bit slower because we have so much more product to work through. And so this dynamic is really tricky because once people get that taste of the incredible acceleration, they're like, well, now it's always going to be like this. It's always going to be this easy. We'll just type it in. But I actually don't see product going away. I think everything still needs to be there. I think it's interesting as you're talking about it and also when I was listening to you on the other show, it got me thinking about this project I did years ago, which was for a friend of mine who runs a commercial real estate kind of education company.

10:00And he wanted to build a product that he saw a need for in that market. And I didn't know anything about commercial real estate at the time, nor do I now, really. But I was like, okay, yeah, I'll help you with this project. And I designed all these screens. And we had a developer team working with us. And it made me think I could have built that prototype now in like a day. But the rest of it, you know, you had to handle some sensitive financial data, things like that. I still couldn't do that on my own, at least. It seems like, okay, the prototype's really accelerated. The rest of it, there's still a lot of complexity and people need to have some deep knowledge.

10:32And now there's things like mythos coming around where it's like there's some serious vulnerabilities in the things we're building. So that stuff's all pretty interesting.

10:39Paul Ford:I want to throw back to you as design leaders, because I think this is really tricky. So I love designers. Sometimes you can replace a designer with a robot that just says we need to do more research before the work can happen, right? And that's the thing I actually like about designers. They're like, hold on a minute. We need to bring it in. We need to understand. And it can drive you crazy when you're pitching a project and trying to get it over the line. But they're right, right? Like we have to get context before this product can get moving. And it's going to start with mood boards and ideation, but it's going to end with us sharing a language with the customer and with the client and with the user.

11:10Paul Ford:And we need to have that language and it's both visual and it's words and it's texture and it's vibes and it's all of that. And what's very wild about this moment and that prototype you could build in a day and the stuff that we build, the language kind of comes for free because LLMs are language machines. And so the common language and understanding that used to take months to understand. And it used to take you a really long time to internalize. People had all kinds of processes and forms they filled out and things that they did. You can get that at the snap of a finger. And what's weird is the person on the other side is like, wow, you really get it.

11:42Paul Ford:You figured it all out. It looks like you did that work. And that is a dangerous one because I think it's very good for closing a deal. Disclosure gets really tricky. I've tried to be really, really transparent with buyers and be like, hey, just so you know, I used AI to help me generate this PowerPoint. I don't really necessarily know what's in this prototype. I sent someone an example from a little like product generator tool I was playing with and they're like, wow, can I have that idea? And I realized I didn't know what that idea was. Like I hadn't looked at it, I just shown them a thing. And I was like, absolutely, I hope you make a billion dollars.

12:17Paul Ford:We're in this very weird world, but I don't really know. Maybe we can pick up the language more quickly and so on. But that part of the work where you're a really good partner long term, where you really understand their domain, where you think about where their revenue comes in and you think about where their success comes in or their clients and their stakeholders. And you build that shared understanding that was always really a critical part of design as a practice. I feel it's under threat. And it's not because the code is really fast. It's because it's almost like we've done a magic trick.

12:47Paul Ford:And it's embarrassing to have to go back and be like, no, that was just a magic trick. I don't really know what your business is. How are clients receiving this idea of, I'm an expert, I know software, I've been at this for decades, and I'm going to go out and use AI to generate this thing for you real quick? Do they feel like, I got gypped. I could have just done that. Or is there like an interesting conversation that unfolds with them? It actually turns out that you need a ton of domain knowledge to pull something across the line. Like there's a reason why 85, 90 % of these projects are failing.

13:18Paul Ford:The problem is that if I didn't really understand the full web stack top to bottom, I don't think I could build you a platform the way I'm doing now so quickly. So I think what's weird is that we learn this stuff as a set of particular techniques and tools. A web person might learn about different approaches to CSS and CSS frameworks like Tailwind. And people learn in terms of JavaScript and they learn in terms of TypeScript. There's this body of knowledge that's very associated with specific technologies. and that is not necessarily relevant for building and learning now. So I think you have to kind of know that stuff just to understand what you're building today.

13:56Paul Ford:But the broader architectural understanding is really hard to pick up because everything has been taught in terms of these specific skills and methods and approaches. So I think it's a struggle. The clients, they're coming to us and what they want, they've heard it can be faster and cheaper. That's sort of where they're coming from. Better doesn't even come into the conversation. I'll tell you why. really big orgs, they can get really great service. They can get Salesforce to stand and sing and dance for them. Small business gets software as a service. They spend 50 bucks a month or 200 bucks a month and that's it.

14:27Paul Ford:Everybody in the middle, which in America gets up to like companies that are making$10 billion a year, is actually kind of stuck. They can't get what they need out of the big platforms. They're too big for SaaS. Their margins are tight. And if you hire a CTO, they're going to want a team and suddenly you're like$20 million in the hole on tech. And so for them, they're going like, wait a minute, I can finally have this thing. We can get off the spreadsheets. Ethel in accounting won't be the only person who can run payroll. Like they're excited, but all they've done is pattern match that AI could get them out of that.

14:58Everybody's triangulating and they're triangulating around cost. I keep thinking about ownership lately as I vibe code my way through a million different products and projects and things that I want to make. And recently, you know, I'm vibe coding, I'm building some stuff, and my GitHub account is just exploding with all these projects. And I'm looking at the git commits, and it's like, has attribution to Claude. And it really burns my butter. Like, wait a second. Okay, I know that you're writing this code, but I'm coming up with the ideas here. And I'm curious about this ownership thing in a client services relationship.

15:36and I can't help but think, because I have a background in painting and drawing, that the art world has already been through this. You take, for example, somebody like Jeff Koons. Listeners, if you don't know who Jeff Koons is, Google it, you will definitely know Jeff Koons, the balloon dogs, the chrysanthemum puppies. Ceramic Michael Jackson. Yeah, ceramic Michael Jackson, all these different things that he would manufacture, which is he would basically create kind of like a factory and say, I wanna make this idea, go make it. But he's the guy with the idea. And of course, the layer behind that is Marcel Duchamp and the famous toilet, the R.Mutt toilet.

16:14And it feels like software is in the R.Mutt moment where we are making the thing. It's a human idea. And the software is a participant, but not the originator. How does this sit in a client relationship or even as an individual software maker?

16:32Paul Ford:I come out of journalism and writing, and we've been through all of it before. We've been devalued before, right? So it's very strange to be in the industry where everybody was the most valuable human being who ever lived and watching them go through that process. Because one thing I learned as a writer is that your work is very rapidly devalued and you don't fully have control over it a lot of the time. And so you have to negotiate where you're going to come down on that. I don't think programmers have ever had that problem. Or if they have, they have. But there's just more fluidity for them.

17:00Paul Ford:And you're right, Claude's name is on most of those commits. Here's what's changing. There's kind of two ways to be a services firm, or there used to be. And one was time and materials, and I'll just bill you for time. And the other is delivery-based. I will get the thing done, and then I will give it to you, and you will pay me in installments for getting it done. And they're very different in a funny way. And we were very much always a delivery-based shop, because what time and materials tends to optimize for long-term is change orders. We got it done, but it's not quite done. I need you to give me a little more money, and so on.

17:31Paul Ford:And if you turn that around, the delivery base is like, yeah, we might have to cut scope here and there, but I promise you it will be done. So now what's funny is that the promise can be kept if somebody is good with these tools. Not everybody is there yet, but I think they will be. Like I think, yeah, maybe only 15 % of AI projects are making it over the line, but it'll be 20 % and it'll be 30 % and it'll get better. And so delivery is much less risky than it used to be. And I think where this ends up and what I'm finding that people want to buy from us is to have us as an outboard auxiliary product team for the long haul.

18:10Paul Ford:So I'm going to pay you X dollars a month and you're going to be in Slack with us and you're going to be using AI. We're going to be using AI, but we're also going to be doing our stuff and shipping our software and there's going to be bugs and so on. And you're just going to be on the team with us. And I think that that relationship gets really, really fluid because what that does is, and what I see people doing, and I think they're getting in trouble this way, they keep trying to use this stuff, but then keep the value for themselves or hide what it did so that it looks like it's their work.

18:41Paul Ford:It's actually an understandable instinct. Like, I don't want you to know how little I did here. I want you just to see and understand that I'm really good at my job. And then when they get caught, it's really, really bad. And I think that's very dangerous. I think you need to disclose when a human did something, when it was produced by us. process, when it's automated, all of that stuff is actually really important. And I also think the buyer is interested to lean in and have that relationship. So I think that's changing, but it's very early days, even though it feels late. I have a slightly different conundrum than Aaron's, which is I've been using the tools a lot for our illustrations.

19:15And I have a little bit of a background in illustration. When I was younger, I did a lot of it. But now we're producing things such a cadence that I can't spend a week working on an illustration necessarily. And we'd also love to have the budget to hire an illustrator and art direct them. But it's just me and Aaron, we're still just trying to get to the point where we can solidly support ourselves. So at some point, maybe we get there. But the conundrum I have is I will art direct something. And for instance, recently, we've been doing these illustrations around Stranger Things, because we had the Stranger Things writer, Paul Dichter, on the show recently.

19:46And so I'll have like Demogorgons in a writing room together or Demogorgon waiter, that kind of thing. It's fun and interesting. And at the same time, I'm seeing the influences it's pulling from. And at one point, I posted one of these illustrations on LinkedIn, I think, and somebody just wrote Maurice Sendak. And I was like, yeah, I'm pretty sure that it hoovered up all of Maurice Sendak's work, and it's drawing from that to create these things. And there's other illustrators out there who have real big problems with them and hate the tools, and that's understandable. But I'm still at the point where I'm creating something, I feel creative, but there is that tension.