Paul Ford: Hi, I’m Paul Ford.

Rich Zide: And I’m Rich Ziade.

Paul: And this is The Aboard Podcast, and today we’re joined by Rafe Colburn.

Rafe Colburn: Hi.

Paul: Well, we’ll get into who he is in a minute, but woof, he understands a lot about how AI and software are coming together. That’s what we talk about here on The Aboard Podcast, so it just makes a lot of sense. Why don’t we play the theme song and get into it?

Rich: Let’s do it.

Paul: All right.

[intro music]

Paul: Rafe!

Rafe: Hello.

Paul: How you doing?

Rafe: I’m doing great.

Paul: You’re coming to us from, now when I take the Brooklyn Bridge, there’s one company that dominates now.

Rafe: [laughing] That’s right. We’ve got a sign.

Paul: You go on over to DUMBO, Down Over the Manhattan Bridge Underpass, and there’s Etsy.

Rich: Overpass.

Paul: Overpass.

Rich: That would be DUMBU, which doesn’t make any sense.

Paul: Thank you. This is going great so far.

Rich: Yes.

Paul: Okay, so let’s get everybody set up. I probably last saw you, like, 15 years ago. You are a long-term technologist and web technologist. And you are the CPTO of Etsy?

Rafe: Yes, I am.

Paul: Okay, so that’s, like, a golden robot in Star Wars.

Rafe: [laughing] Maybe.

Paul: What is that job? People may not even know what those letters all mean together.

Rafe: Sure. So I’m the Chief Product and Technology Officer. Originally started there in engineering, but I’m managing the product managers and designers as well now.

Paul: Okay, so how, how big, like, your team’s big, right?

Rafe: Large. It’s over a thousand people.

Paul: Ooh, okay. Wow. Okay. How’s that, how’s that feel to you?

Rich: You put them into VP buckets.

Paul: [laughing] It’s true.

Rich: That’s how you keep it organized.

Paul: So big. Mostly in New York City still, or pretty spread out?

Rafe: Some of both. I think, you know, like a lot of companies, a lot of people moved out of New York in the pandemic and so, yeah, pretty split.

Paul: Okay.

Rafe: A lot of remotes.

Paul: You’ve been there for a while, but the job is relatively new, right?

Rafe: Yeah. So I started at Etsy as an IC engineer in 2012, and then I took, like, three years, three or four years in London, I was actually the CPTO of Depop, which Etsy owns, clothing resale site.

Paul: Mmm!

Rafe: Kids like it. Yeah, and then I came back as CTO, and eventually CPTO, just, like, in May. So relatively recently.

Paul: Okay, great. So we want to talk about two things with you. One is AI. Just…AI. [laughter] No, no. But also in particular, how you see your role with a thousand engineers evolving, because there’s a lot of robots that say they can write code now.

Rafe: Yup.

Paul: And a lot of people writing code, and sort of how you’re metabolizing that change into the org and sort of what other CTOs should do. So that’s one. And then two is you guys just made a deal with this really nice company, OpenAI, ChatGPT, they’re good, they’re, you know, just little guys, little guys helping them get a leg up, where all of your stuff, if I search and do shopping searches on ChatGPT, which I actually do, Etsy stuff just gets in there. And I want to talk about that integration, because it seems like you solved the generative engine optimization problem the smart way by just making a big old deal with the LLM itself. Do you have any other questions?

Rich: Yeah, I think it’s worth it to take two minutes and talk about what’s unique about Etsy’s engineering culture. And I think it’s been unique for a long time. I’ve known about it because I’ve known people who work there and whatnot. But I think if you take a couple of minutes and share what’s distinct about it.

Paul: It’s worth noting, too, I mean, it’s a big New York City engineering culture, which is unusual. We don’t have a lot of—

Rich: We don’t. We don’t.

Paul: —shops at this scale that are centered here. So…

Rich: And this uniqueness predates AI or anything like that. So if you could talk about that for first, sets us up nicely.

Paul: Is it all still in PHP?

Rafe: A shockingly large amount, yes. I mean, of course we have mobile apps that are in the right mobile app languages and things that are search stack and ML stack. You know, there’s more Python and things that run the JVM Scala and so forth. But the actual website and the entire backend are basically still a giant PHP app. Yeah, that is still true.

And, I think, going to Rich’s question, you know, it’s been interesting. So of course I’ve seen, not quite since the beginning, but close to the beginning, you know, if you go back to, like, 2010 or something like that, they had kind of a regime change in the technology team. And the Etsy engineering culture is in some ways a mutation of the Flickr engineering culture from way, way back.

Rich: Mmm.

Paul: Mmm.

Rafe: Which was, the kind of core tenets of it were massive observability, really simple stack, continuous deployment back before everybody did it, and branching in code. So feature flags, no, no, you know, working offline and then you just, like, get the code into production and hide it behind a feature flag, you know, which I would say, one of the reasons I went to work at Etsy in the first place was because this was just revolutionary at the time, and no one was doing it that way.

Rich: Yeah.

Rafe: And so I was like, I want to kind of see it up close. I almost didn’t believe it would work. I wanted to see it myself. It’s one of those things that now if you’re starting a startup, you’d be like, “Yeah, of course we do all that stuff.” But I think at the time probably we were very early into really using A/B testing to test lots of things, especially for a company our size.

And you know, so I think that kind of data-driven, high-velocity, continuous deployment on top of a very operationally simple stack was, you know, was the kind of hallmark of the culture and I think still is. And you know, I would say a lot of things flow from that. You know, like I always used to say we’re more agile than agile, because if I have an idea, I can get it into production today, so I don’t need to wait for a sprint. Like, you can get it out whenever you need to, I think—

Paul: So wait, that’s a nice thing for an organization to say. Everybody likes sort of real agility and we can deploy.

Rafe: Yup.

Paul: How do you maintain that? You got a thousand engineers. They can’t all be—and actually we should just take a break, for all the youth in the audience, Flickr was a photo-sharing website.

Rafe: Yes.

Rich: Yeah.

Paul: And a lot of the people in it went to Etsy and then kind of went to, a lot of them went to Slack. Like, there’s a kind of…

Rafe: [laughing] That was the great divide. I think half the leadership founded Slack and then, you know, folks, the other folks came to Etsy.

Paul: But Flickr also really represented kind of Web 2.0, like, active applications with a real specific forward velocity. And so it was a big cultural change.

Rafe: Yeah.

Paul: And so when you’re talking about that, that’s kind of what you’re talking about. All right, so now these are all really nice ideas. Everybody loves them. But very large organizations can’t ship anything quickly.

Rafe: Yeah, yeah, yeah.

Paul: So you’re saying it’s maintained. You’re saying that you can still do stuff fast at Etsy.

Rafe: Yep.

Paul: Okay, how? How do you keep that going?

Rafe: Yeah, it’s a really good question. And you know, I think we’ve been, like, at the precipice sometimes [laughter] where it’s like, “Oh, we’ve scaled up and like pushing code has gotten slow.” And I think there’s just been this real kind of top-to-bottom commitment to maintaining all the infrastructure and developer tooling that allows us to go fast. You know, and because that’s kind of never been sacrificed, we are still able to do it even a much larger scale. We’ve had to change the way things work here and there, but, you know, we’re still deploying code to production dozens of times a day. And you know, the app goes out, weekly releases, we almost never miss. And so it’s just, you know, that’s—and that’s why, you know, I think that’s, you know, they say culture trumps process or whatever. I think the culture is, you know, we, you know, we’re committed to this kind of way of working. And so…

Paul: So back in the day, the thing Etsy was famous for is if you start there as an engineer, on day one, you’re shipping some code.

Rafe: Yeah.

Paul: Is that still going on?

Rafe: Sadly, no. [laughter]

Paul: Fair! Fair. That’s, like, 20-people team, not 1,000-person.

Rafe: But I’d like to get back to it. I think it’s totally doable. I just think it’s a habit we got out of more than a, like, it’s not infeasible, it’s just changes.

Paul: Talk a little bit about your own career, too. Like, you start as an IC.

Rafe: Yep.

Paul: It’s a website that sells stuff that people make.

Rafe: Yep.

Paul: And it’s got a nice, dynamic culture. It’s a little crunchy. Very pleasant. I’ve been to the offices, you know.

Rafe: Yep.

Paul: Now you manage a thousand people.

Rafe: Yeah.

Paul: Okay. So I think the differences in those two worlds are pretty obvious. But what steps along the way, where did you actually have to really go learn something new to get away from engineering and towards humans?

Rafe: Yeah, I was actually really lucky. I moved into management just a few weeks after I started at Etsy, and all the managers were there, at the time, were kind of learning to be managers.

Rich: Mmm.

Rafe: You know? And so it’s a little bit of a community of practice. You know, we had a guy there who, who ran engineering, this guy, VP of Engineering, and he just, like, really, really believed in management as a real practice that you try to get good at. I would say this is probably like peak Rands in Repose era, where, like, everybody just woke up and realized engineering management is a real job. And so we all—

Paul: It is a deep blog cut. [laughter] Again, for the children, but, like, there used to be these blogs where people talked about engineering management all day long.

Rafe: Yes.

Paul: And they were big and important because—I think because the industry was maturing, lots of new people in and nobody knew what to do with all these humans.

Rafe: Yeah. And I’d been an engineer prior, I mean, an engineering manager prior. And you know, my basic approach was I did engineering myself and I told people what to do when I needed to, but mostly left them alone, because that’s, like, what engineering was. I didn’t even know what a one on one was.

Paul: Mmm hmm.

Rafe: You know? And I think that was just kind of how the tech industry was—probably, maybe all industries back then. And so, you know, we sort of all went on this journey together. I was lucky to be in this journey when a lot of people were on this journey where, where, like, management as a craft became a thing people took seriously. So, you know, and I mean, obviously a ton of stuff to learn, you know, I think probably made so many bad mistakes as a manager early on. [laughing] You know, where I could have given people more useful feedback and I could have, you know, I surprised people with negative feedback and did lots of things that were not good. Didn’t know how to talk to people about compensation and all those things they care about and are really important to them. So, you know, you just kind of get the reps in and you know, embarrass yourself a few times and eventually you get there.

Paul: Still working on it, man. Over here.

Rich: Mmm. We’re good at it though.

Paul: We are. [laughing] We’re good at the humiliation.

Rich: Shout out to our happy employees.

Paul: Oh yeah.

Rich: Yes.

Paul: Look at them out there, smiling [laughter] and just looking in the window, waving. What do you got? I have a lot of questions, but…

Rich: I mean it’s, it’s a big org, your group.

Rafe: Yeah.

Rich: Your group, groups really. It’s worth noting, it’s not just engineers now, it’s product people, it’s designers. Yeah, I mean, let’s get right into it. I mean, you talk about a culture that gets out of your way and lets you be productive, and there’s this new technology called AI that gets out of your way and is supposed to help you be productive.

Rafe: Yeah.

Rich: There’s a lot of anxiety. We’re seeing a lot of anxiety out there around what this technology and what these tools mean—not just for orgs, I think the orgs, they take a minute, they’re sort of slower to realize the change, collectively. But individuals. I guess, what are you seeing? You’re at a particular scale and—

Rafe: Yeah.

Paul: But also in a very high-velocity-oriented organization. And these are velocity-promising tools.

Rafe: Yeah.

Rich: Yeah. So how are you thinking about it? What are you seeing? What would you tell the audience, even?

Rafe: I mean, it’s kind of a crazy time because you have people sort of bottom-up, people are afraid about their job changing. Maybe losing their job, maybe just not getting to do the things that they like about their job as much as they used to. For the most part, engineers follow the technology-adoption cycle same as everyone. And like, most of them are not early adopters. They’re, like, mainstream adopters, right? And so, you know, I think—and the adoption curve is going really fast here, which is scary to people who are not early adopters.

So there’s that, this huge amount of anxiety on the ground, and then there’s this completely other anxiety which is, like, at the sort of executive level, at the top of the business level, people are desperately afraid of being left behind. And the hype is crazy. And so I think, you know, it’s like, “Oh my God, you know, are our competitors doing this thing that we should be doing? And if we’re not doing it doesn’t mean we’re not going to succeed and we’re going to lose later?” And so when you have these two things come together, it just, like, it gets crazy on the internet, right? [laughing] That’s what we’re seeing.

Rich: Yeah, yeah.

Paul: Well, I gotta imagine too, like, so much of the—like, so much of generative stuff, it’s a little bit like, “Well, we’ll see how this shakes out.” But so much of the code gen and Cursor and so, like, that’s starting to really lock in.

Rafe: Yeah, yeah, yeah.

Paul: So it feels like our world is one that’s getting hit with the hammer the hardest.

Rafe: Yeah.

Paul: It’s the same amount of hype, but it’s actually sort of real change.

Rafe: So two things are true. I think one is very few companies don’t succeed because they cannot get enough code into production—before or after AI. [laughter] Like, “Oh my God, the engineers are just not writing and deploying enough code. That turns out to be what’s holding us back.”

Rich: Yeah, yeah.

Rafe: And so I think it’s almost always strategy prioritization and some of these other things. So, you know, I think some ways that whole productivity discourse is maybe a little bit misplaced. And then I think the other thing that’s true, I’ve been thinking about this a lot, is in some ways, like, the other thing about the rise of open source was it was sort of a conspiracy by the ground, a ground-up conspiracy. Management didn’t want it. And I think if no one told anything in the executive suite about AI, engineers would think this is the greatest thing that ever happened to them in their entire career. They discovered it themselves. It is so useful for so many things.

Paul: Well, notice how all of the effort has gone in, like, the most ironic way into developing development tooling.

Rafe: Right.

Paul: Like every AI startup that’s worth a lot, essentially the same as configuring your text editor, but just at like a $6-billion scale.

Rafe: Yeah, yeah.

Paul: Cursor and Replit and so on. Is AI checking in code at Etsy right now?

Rafe: No.

Paul: Okay.

Rafe: No.

Paul: Have you set a policy for how things are supposed to go?

Rafe: No.

Paul: Okay.

Rafe: But what I have tried to do is give people access to the tools they need and then kind of rely on the early adopters and early majority folks to lead the way where we’re going. So we have, like, an agentic coding Slack channel. It has hundreds of people in it. There are people in there sharing what they’re doing. They’re trying all the new stuff that people are writing about on blogs and comparing. And so I think probably, you know, I think there’s just so much innovation going on inside Etsy, there’s, I mean, outside Etsy, there’s innovation internally about how we can apply it to Etsy problems. And so I just kind of want to step back and let the innovation happen a little bit and then we’ll figure out what direction we’re going to go in, rather than trying to be directive, I just don’t think I’ll get it right.

Paul: I think that’s, one thing we talk a lot about with people is like, don’t try to capture all the value for yourself and then just hand this AI thing over to people. But actually it needs to be this conversation in order for it to work. So it’s—okay, so that one Slack channel is where everybody is in a safe place. They can go in, they can be like, “I’m using these tools, I’m doing stuff with it and I’m learning these things.”

Rafe: Yeah. And I think ultimately, you know, the policy at the ground, at the bottom is like, you’re responsible for the things you check in to GitHub, you’re responsible for the emails you send to people, you’re responsible for what’s in Google Docs. If you use AI to make it and it’s better, great. If it’s, use AI and it saves time, great. But ultimately it’s you, not—

Rich: They’re your words.

Rafe: AI. Right.

Rich: It’s your code.

Rafe: It’s your code. Yes.

Paul: Yeah, it’s just like, you know, Google gets in trouble if a Waymo does something bad. Right?

Rafe: Yeah.

Paul: Like, it’s just—okay, so it’s, okay, so humans get to use the tools, but they have to take responsibility. Which means they have to read the code it produces.

Rafe: Yeah, of course. But you know, maybe one thing, I think the, like, “read the code.” Totally agree people need to do it. But, like, I’ve been around long enough to see people copy and paste things from Stack Overflow that they had no idea whatsoever, or things they find elsewhere in the code base. And like, do they read the code? No, they just get it to work and say the bug is fixed.

Rich: Yeah.

Rafe: And so not reading the code, I think, it’s a human problem, not an AI problem. [laughing]

Rich: Yeah. You’re at the executive level. Insert dramatic music here.

Paul: [singing dramatic music]

Rich: Let me be a naive executive at Etsy for a second. And I walk up to you, Rafe, I’m like, “Rafe! 1,000. Should be, like, 600, no?”

Paul: “Or maybe 2!”

Rich: “This magical technology.”

Paul: Yeah.

Rich: I mean that sort of anecdotal, sort of casual sentiment is taking hold in people, you know, across the aisle at companies.

Paul: We hear about it more in, like, finance firms and less in tech.

Rich: Yeah. But you know, there’s non-technical executives at Etsy. Right?

Rafe: Sure.

Rich: And they’re reading, they’re reading the stuff and they’re seeing CNBC and all that.

Paul: You’re gonna make Rafe make a case for humans?

Rich: No, no, he may make a case not for humans.

Rafe: Yeah.

Rich: Who knows what case he’s going to make?

Rafe: Who knows what I’ll say! [laughter]

Rich: This is what makes this exciting, right? I mean, that is a sentiment that’s out there.

Rafe: Sure.

Rich: But at the same time, they’re also terrified at the risks of turning a lot of things over to machines. So speak to that for a second. I mean, that sentiment is real and it’s there. We’re seeing it everywhere.

Rafe: Yeah, I mean, I guess this is where having a long history in the industry really helps.

Rich: Yeah.

Rafe: Because just as I said, few companies don’t succeed because they don’t check in enough code. The alternate side of that is also true, which is like, could we make more use of more software? Absolutely. I think, do we have great internal tools? No. Why? Because it’s not economically feasible to write great internal tools with a thousand people. But if AI makes it feasible for us to build more of our own stuff really effectively, we would probably do more of it.

So, and I think going through history, productivity tools are not new to developers. When I started my career, we paid so many people to write HTML by hand. Like, you just had to write, you know, tags and, and like the industry did not go away and we didn’t have less people working in tech now than we did when it became possible to like have all your HTML generated. Right? Instead we have many more websites and a lot more technology and a lot more software. And so I’ll worry when people are like, “Boy, we just, we have so much software, I guess we just don’t need any more.”

Rich: Yeah.

Rafe: But, yeah, I don’t think we’re close to that. [laughing]

Rich: Yeah, I mean, my observation is I think companies fall into two categories: Ones that view technology as truly additive and differentiating, and others that view it as a department that is, that is like, they lump it in with IT services, right? And I think the ones that lump it in with IT services, it’s like, “Go get me 10% more margin. Why not?”

Rafe: Yeah, sure.

Rich: The cool tools are here. Does Etsy consider itself a technology company?

Rafe: Yeah, unquestionably so.

Rich: Yeah. And I think that obviously frames the conversation as much as anything. Okay, next naive executive across the hall comes up to you and says, “Why doesn’t it do more magical things? Like, my daughter just showed me ChatGPT, do all kinds of smart stuff?” How is that manifesting itself? Or like, how does that seep into the roadmap? And I’m guessing you’re getting strange asks.

Rafe: Well, I mean, we’re a technology company, so not too strange from our executives. [laughter] That’s maybe another advantage of being a technology company. Like, probably, you know, my number-one push for any executive is if you’re talking about AI all the time and you’re not using it, you know, then, then what are you doing? You know, you expect your employees to automate their way out of their jobs or to do these other things?

Rich: Yeah.

Rafe: But it’s like, how are you, you know, are you really getting a feel for it? I remember I read on a weblog recently, the deceptive thing is, like, people think using AI chatbots, things like that, coding agents is really simple, but it’s not simple. The surface area, of course, incredibly simple. You can type into a box and it will tell you something useful.

Rich: Yeah.

Rafe: But I think whether you’re going to use it to help you write something or going to use it to help you write code or do all these other things, like, expertise and going deep really, really matters. And so I think part of the problem is, you know, I think it’s a power tool. I would say to any individual contributor, like, hey, if you can figure out how to really be an expert and harness AI to do your job? You will have a massive competitive advantage over your peers. You know, but if you think you can just, like, type stuff into a box and make your job go away, you just kind of miss the point. And I think a lot of executives are probably still in the, like, you type into the box instead of doing real work and you make money and it’s just not true. So I would say, like, do your work.

Rich: I mean, that’s good advice for the individual contributor, even, who thinks they can just work an hour and then go chill for the rest of the day.

Paul: I have—so Claude gave me $1,000 in free Claude code credits, because I’m on one of the plans, because they want us to try the web-based tool. And I mean, I have been doing my best to spend down those credits. I can’t. Like, I just don’t even know what to do. I’ve got it parallelizing. I had to just, like, document an enormous code base because I thought it was interesting. Cost $2.

Rafe: Yeah, yeah.

Paul: It’s just, but that thing that you just described, even if you spend a lot of time with these technologies, and even if, like, we’re building a business kind of around them, it’s a lot harder than people think. Like, waving your arms and saying, “Let’s go agents,” is really different from actually implementing anything and getting value out of it.

Rich: Yeah, I mean, there’s studies coming out today about productivity not showing up yet. They’re spending all this money because there’s, it’s, we’re in like the panic part of the hype cycle right now.

Rafe: Sure.

Rich: It’s like, “You must buy it. Buy the thing.” And I think they’re just parachuting it into large orgs and they don’t know how to make use of it yet and whatnot.

Rafe: Yeah, I mean, it’s just, you know, speaking of long traditions in our industry, you know, clueless companies getting worked by vendors is one of the, like, oldest in the world.

Rich: Shhh.

Rafe: You make a hype cycle, and—

Rich: Rafe.

Rafe: [laughing] Sorry, not you, not you guys.

Rich: Rafe, not us. Not us, man.

Paul: Remember where you are, my friend. [laughter] So recently ChatGPT rolled out a thing we were all craving: ChatGPT shopping.

Rich: Is it a thing, or is it just part of the box, the magic box?

Paul: Don’t ask. That is an epistemological question that can’t be answered today.

Rich: Okay, fair.

Paul: But anyway, when you type things into ChatGPT and say, “Help me buy something,” it’ll throw up a bunch of products. And actually it’s a good product. I used it, I wrote about it at one point.

Rich: It’s visual. It tiles out the products with prices and links and all that.

Paul: And it’s a strong argument for what people have been saying about GEO, which is terrible, but it’s generative engine optimization instead of search engine optimization, where you want to get your stuff into the LLM so that they can tell people about your stuff.

So people are obviously very interested in this. And Etsy was one of a few people—a few large companies that ended up providing all of their stuff to this new world and it appears in the search results. So I think just talk a little bit about that. How does that work? What’s it, what’s that relationship like?

Rafe: Yeah, absolutely. So I think under the hood it’s, in a way it’s a little bit like search ads or something in that we provide them with a feed. So rather than just being scraped or whatever else, we actually have a product that we sent to them.

Paul: Okay, so there’s an endpoint and they can just go get all of Etsy’s data. Not all of it, but, like, the product data.

Rafe: Not quite. We send to them.

Paul: Oh, okay.

Rafe: It’s push. It’s push. It’s almost like you use with, like, Google Shopping or something.

Paul: So they have an API for you.

Rafe: Yeah, yeah.

Paul: Okay.

Rafe: Yeah. Batch.

Paul: Yeah.

Rafe: And then from there basically you know that we also have a checkout integration so that people can buy things on Etsy by way of OpenAI rather than having to, like, link out. That was, I think we were the first company to—

Rich: Oh, wait. So you could buy inside of ChatGPT?

Rafe: Inside ChatGPT. So ChatGPT kind of just purely with scraped information, will show you the product cards. So you get into a conversation with it and it says, “Do you want me to show you four bird feeders that might suit your requirements?” And you say, “I would love that, ChatGPT.” And it shows you some product cards in addition to description of items.

Paul: And actually let’s go back one step which is, people won’t know this, there are open schemas for how to describe products all over the internet, which really Google kind of pushes now.

Rafe: Yeah.

Paul: To get e-commerce kind of baked into—

Rich: Good sort of metadata, yeah.

Paul: Yeah. Into search. I’m guessing Etsy is, like, all over that, right? There’s always, like, all those pages have rich metadata. So that’s a world where Google would come and, like, spider your stuff and be like, “Hey, thanks, we now know all about this product.”

Rafe: No, we have, like, a huge streaming integration with Google.

Paul: Okay.

Rafe: So, like, so like we have a bunch of infrastructure that sends up-to-date pricing, inventory, updated descriptions, just like, so it’s, like, more like an internal search indexing pipeline. So if you’re, like, using Google Shopping or you see Google search ads, yeah, we provide them with all that information. It’s not, it’s not from scraping.

Paul: So you push the fire hose towards—

Rafe: Yes.

Paul: Like, everybody else might be, it might be scraped. But, like, a big org like Etsy is going to—

Rafe: Yeah, I mean and in MarTech world there are a lot of little, like, SaaS companies that will do that even if you’re a small company. Probably, like, one thing, like, a WIX or a Shopify will, you know, they’ll provide a Google shopping integration.

Rich: Sure.

Rafe: So your products get syndicated. Like, it’s just product syndication.

Paul: It’s a checkbox.

Rafe: Yeah.

Paul: So that’s where that whole world lives.

Rafe: Yes.

Paul: And then—which Google probably really prefers because it’s verified if it’s coming from you.

Rafe: Exactly, exactly.

Paul: Okay, so now we’re going to take that same fire hose and point it towards OpenAI.

Rafe: Exactly.

Paul: Okay, so it wasn’t that big a deal.

Rafe: Well, I mean it’s the work but, but it builds on existing concepts. [laughter]

Paul: No, I guess, like, it wasn’t—I’m sure it took a lot of effort and time and relationship building.

Rafe: Yes.

Paul: But conceptually, it’s the same idea.

Rafe: Conceptually, the same idea.

Paul: Okay.

Rafe: Probably, like, the easiest way to think about it—I don’t work at OpenAI. I don’t know how their stuff works but you know—

Paul: Neither do they. [laughter]

Rafe: Yeah, I mean they are growing fast. I think, you know, they have a kind of traditional product search that they use and then you can kind of think of the LLM probably as like the final pass on that search. So they get some set of candidates. Like, the LLM can’t, you know, they, I’m sure they have billions of products. They have tens of billions from us. You know, and so you can’t, an LLM can’t search all those things. It’s too much context, right? So they narrow it down using some other old-school technology. Then they pass it to the LLM to really, like, narrow it to the few items.

Paul: Where do you see that all going? Is that, I mean it feels like, I mean if you leaned in on this, it seems—Etsy has clearly decided that LLM-based search technologies are going to be part of its future. Okay? So that’s, that’s A. But, like, if you’re the CTO of an e-commerce company that’s a lot smaller, right? What do you do? Are you just going to wait for OpenAI to say, “Okay, now,” or do you try to get in there? Like what’s, I don’t know, how’s this going to go?

Rafe: So I think your stuff is going to show up no matter what.

Paul: Mmm hmm.

Rafe: Because they do spider everything these days, or scrape it or whatever.

Rich: If it’s on the web, it’s going to get slurped up, yeah.

Rafe: Exactly. And I think probably if you’re on some e-commerce provider, they will build some kind of OpenAI integration that you can, again, it’ll be a click box, a box you can click a little bit like, a little bit like Google. So I think that’s probably where we’re headed for the short term. If you’re building your own thing, like, I’d imagine the OpenAI partnership team is very busy. I don’t know if you’re a small business if they would even return your call.

Paul: Yeah, yeah. Actually on that, on that note, and I don’t want your PR person to jump across the table, but like, how’s that going? Like, how’s the, the integration going? I know you probably can’t tell me, “Hey, we’re making lots of money off of, off of OpenAI,” but does it like, does it work? Is it a thing?

Rafe: I mean it works, for sure. It works. And I think, you know, we are seeing that the customers that we acquire that way are valuable. You know, I think in part because—

Rich: It’s kind of reciprocal, in a lot of ways.

Rafe: Yeah, exactly. And I think the reason for that, and maybe when I think more broadly about our strategy around LLMs, whether it’s, or generative AI or whatever we want to call it, whether it’s internally or externally, is, you know, I think the hard part for Etsy over a long period of time has been we don’t have the metadata that, like, an Amazon has, for example. We have, like, millions of small sellers. They just describe their items.

Paul: People have to type it in. Right, yeah.

Rafe: People have to type it in, they have to do it right. Like, even Amazon struggles with metadata. You can imagine how it is for us. But to an LLM, everything is metadata. Everything is an API.

Paul: Catalog search, and sort of all the metadata stuff that you need to do to make a…and the taxonomy stuff can actually be solved now. Whereas asking humans to do it is…a mess.

Rafe: Yeah, it’s a nice augmentation. And then I think also, you know, it can take all that context, whether it comes from images or text or anything else, and if it knows what you’re looking for, can write you the sentence you really care about. Like, if you go to a listing on Etsy, you know, there’s a description and there’s a shop about page and there’s the photos and everything. And so you might say, “I’m looking for a gift for my mother. You know, she loves her, she loves her, you know, Goldendoodle,” and all these other things. Then if it’s going to pull up five listings from Etsy, it’s going to tell you exactly what about them would appeal to a goldendoodle enthusiast in a way that would actually require a lot of research on your own. And so it’s like a, it’s just sort of, like, a context synthesizing shortcut, which is really beneficial to Etsy, because I think we have a lot deeper context on things than most sites have. So making it more accessible is like a huge win.

Rich: I want to ask you about the future, Rafe.

Rafe: All right, love that.

Rich: And I’m going to run a scenario and I want you to tell me A) if we’re going to get there and B) when.

Rafe: [laughing] Okay.

Rich: Ready?

Rafe: I’m ready.

Rich: All right. I type in “blue Crocs that have cool yellow streaks on them.” And the first thing it does—

Paul: That’s a very realistic query for you.

Rich: Zing. It shows me an image that it generated.

Rafe: Yeah.

Rich: And it puts a price on it that it generated. Actually shows me a few of them.

Rafe: Yeah.

Rich: Variations on it. And I pick one of them, and then when I put it in my cart and purchase it, it actually generates the product and sends it to me.

Rafe: I’ll say 100% feasible. And I think soon, if you’re, you know, depending on the brand. I think Crocs is really interesting example because I feel like Crocs is like, I mean, they’re just these little injection-molded things, right?

Rich: They’re ridiculous. And people put chicken fingers, toys, on in the holes. Have you seen this?

Rafe: They’re called Jibbitz.

Paul: Yeah.

Rich: Jibbitz??

Paul: Yeah. It’s, like, a whole thing. It’s for kids. You’ve never done this.

Rafe: And adults, too. Don’t kid yourself.

Paul: Oh, really?

Rafe: Yeah.

Rich: Jibbitz? It’s called?

Paul: Jibbitz.

Rich: There’s, like, little French fries and chicken nuggets. I’m like, “What is going on?” I have preteens. And I was like, “What the heck’s going on?”

Rafe: They’re like the new buttons on your denim jacket.

Rich: Yeah. Oh, okay.

Paul: We’ve gone down a path here.

Rich: I mean, the day, the days of asking for a Jibbitz on Etsy that doesn’t exist, that isn’t in your product catalog, and it just gets generated in response, feels utterly feasible and feels not far away.

Rafe: I think it’s utterly feasible and not that far away. And I think especially if you’re willing to, like, you know what I mean, right, like, 3D printing is real. Crocs doesn’t even need 3D printing. They just need to put the right color foam into the machine, you know, and I think—

Rich: But not, what about anything, a sweater, a particular iPhone case that has glitter on it. All these things, like, on-demand production.

Paul: Let’s not, like, Etsy probably won’t be solving that problem, but the problem that Etsy will have is that there’s a combinatorial explosion of options.

Rafe: Right.

Paul: That are related to any of those kind of product offerings. Right?

Rafe: Yep, yep.

Paul: How do you handle that one? Because literally, Crocs could show up tomorrow and be like, “Hey, we love you guys,” and over 2 billion different ways of generating Crocs are now, “Here. I made you 2 billion pictures.”

Rafe: Yeah, yeah, yeah.

Paul: Which is, I’m saying that about Crocs, but literally someone in their room could be doing that right now, and they can flood you.

Rafe: Uh…

Rich: I don’t think they’re gonna go to Etsy for that.

Rafe: I don’t think you can sign up on… You know, like, Crocs is definitely not within our policies as a seller.

Paul: Okay. Yeah.

Rich: That’s, I think they’d have to, there is also, by the way, you can customize a sneaker on some sites. Just takes a month or whatever it is. Are we gonna eat away, I’m trying to kind of draw this AI dystopian creators are just sort of left in test tubes and they don’t get to actually do anything anymore. And I feel like on-demand creation—I’m thinking, this idea kind of got sparked by just image creation, which is horrible and slop and all that.

Rafe: Yeah, of course.

Rich: But can this transition into products and into creative, like, what were otherwise creative?

Rafe: I mean, it’s an interesting question that I don’t think a whole lot about. I mean, I guess the big bet on Etsy is obviously human creativity is going to remain important. We’re all in on human creativity and human connection.

Rich: I mean, yeah. [laughs]

Rafe: That’s our differentiator as a brand.

Rich: Yeah.

Rafe: And I think we very sincerely believe in it. And I think that’s true. I think, look, most people are not good at making and designing things, and it’s not because they’re not good at using the tools. It may be in part that, but also, if I said, “Boy, I don’t have really amazing furniture in my house,” I could learn all the carpentry skills in the world and I still wouldn’t have amazing furniture because I’m not a furniture designer. [laughing]

Paul: It’s actually analogous to what you said around the problem not being the software, that we don’t have enough software.

Rich: People focus on the tech as this thing coming for you.

Paul: Rafe, your point of view seems to be, like, just because people can, they will, but it doesn’t mean it’ll be good or other people will want it.

Rafe: [laughing] I think that’s right.

Paul: That’s true of the engineering and that’s true of the products and so on.

Rich: I want to end it with some advice you can share, as someone that manages a lot of different people and a lot of different disciplines and whatnot. Young people are also nervous. They’re trying to, you know, they’ve made particular bets on their careers and… Advice to the young engineer, the young product manager, the young designer who’s watching all this change come at them.

Rafe: You know, I would say really lean in. I think that’s my advice. If I think about my early career, one of the big problems was it was just hard to have the, you didn’t have the tools to build real things. You know, kind of especially, I started right when the web started and like, tooling was really expensive and hard to come by, you know, and, you know, you couldn’t, you know, and so I think now—

Paul: For people who don’t know, like, a good example would be like a database. Like, it was hard.

Rafe: Yeah. Everyone’s—

Paul: Yeah, you had, it was expensive—

Rich: A web server.

Paul: A web server, right. These things took hours and hours and a lot of money to set up.

Rafe: Yeah, exactly, exactly. You know, I mean, I think even, yeah, before Linux was ubiquitous, if you want to use Unix, like, the machines were just incredibly expensive. You know, now we have this amazing ability, I mean, you know, I like it, I like building things with coding agents because I can like, you know, produce copious amounts of code and create functionality quickly and, you know, kind of in my spare time. But I think if you’re, if you’re new to this industry, like, man, being able to build whole products and see how they work, you know, I think is really, really cool. Like, I think I would have loved that. You know, it’s just so hard to build those things.

And so if I go back to the beginning of my career, you know, I started about the time the web kicked off when, you know, and so I knew a lot about the internet from having used it at university and about those things. And I think I was just able to adapt so much more quickly than people who are, like, you know, aging enterprise developers who use like, PowerBuilder and Visual Basic.

Rich: Yeah.

Rafe: And I think, you know, you have like that, whatever, mental plasticity or whatever, what they call it, like, lean into it and show how you can outperform people who’ve been doing it for longer.

Rich: I think, I think this is, I think what you’re saying is really interesting here. I think without knowing it, I’m early internet, too, Paul’s early internet, too. I mean I came out of law school and I was like, “This internet’s too interesting. I’m not going to practice the law.” I think it moved at a pace where you internalized how it worked and you actually, it wasn’t a 101, it wasn’t a particular certification. You just kind of got a sense of how it worked.

Rafe: 100%.

Rich: And it’s, this is first off, the speed at which this has all landed is incredible. And so for a young person taking the time to understand it, which you don’t have to, I think that’s one of the things here. It’s such a consumer experience right now that you could just use it and spew out code and see things light up and whatnot. You don’t have to bother trying to understand it. But I think you’re right. I think understanding it is how you get your angle and really appreciating what it’s about, which many people won’t bother with. Right? Because it’s easy.

Rafe: If you use it and you don’t understand it, you might be able to outperform someone who doesn’t understand it at all.

Rich: Yeah.

Rafe: But—or doesn’t use it at all.

Rich: Yeah.

Rafe: But are you going to be able to outperform someone who really understands it?

Rich: That’s right.

Rafe: And so what happens when you, you are the 60 or 70% of people really understand it and you’re like, “I just kind of mail it in. It’s just like any other job.” [laughter] You’re just going to be, like, bad at it. And so, so yeah, I mean, I guess in the land of the blind, the one eyed man is king. But, like, it’s not going to all be one eyed man forever. Right?

Rich: Yeah.

Rafe: I think you have to think about that, right?

Rich: Right.

Paul: Awesome. So look, if you want to talk to you, how do they get in touch?

Rafe: Probably LinkedIn is the easiest. Yeah.

Paul: For all of us now. That’s what happens.

Rafe: Just type my name and get a message box. That’s about as good as it gets.

Paul: Well, I don’t, I have about 2 billion more questions, but I think we have to leave it there just for time and hopefully we can follow up with you at some point.

Rafe: Yeah, absolutely. Yeah. Really fun. Thanks.

Paul: Rafe, it was great to have you here. Thank you for coming in.

Rafe: Of course. Of course.

Paul: It sounds like a nice, healthy scaling engineering culture, which is a lovely thing to see.

Rich: Jibbitz.

Paul: [laughing] Jibbitz.

Rich: Thank you, Rafe. This was a lot of fun.

Paul: Friends. If you don’t have a thousand-person, incredibly well-run large operating engineering culture, but still have a lot of technology that you need to build, where could you go, Rich? What could you do?

Rich: Oh, the way you set that up for me, Paul. Appreciate it.

Paul: It’s my English degree.

Rich: Aboard.com. We stand up enterprise-grade software really, really quickly. People are nearby who will tease out exactly what you need, work through it, but at a fraction of the time and cost.

Paul: Nimble, thoughtful people who understand how AI works.

Rich: Go play with it yourself at aboard.com. Like and subscribe. We’ve been told to say that.

Paul: Five stars. Ring the bell. All that stuff. We need that.

Rich: Like and subscribe.

Paul: It’s very organic and natural the way we do it.

Rich: Yeah.

Paul: All right, Rafe, thank you for coming in.

Rich: Thanks, Rafe.

Rafe: Of course.

Rich: Have a good week, everyone.

Paul: Bye, everybody.

[outro music]