Stack Overflow’s Pivot with Jody Bailey, CTO

(Intro Narrator at 00:00:00) Today, we're talking to Jody Bailey, CTO of Stack Overflow, about how they're rebuilding and revitalizing the platform. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:17) So the theme that we have is rebuilding and redeeming Stack Overflow. I was very excited about this conversation, partly because I'm a software engineer of over twenty years. Stack Overflow was a huge part of my growth and development. I learned so much from it. And then on the podcast last week, we were talking with the CTO, and we're talking about AI and generative AI and all of that.

(Joel Beasley at 00:00:40) And we're like, what's going to happen to Stack Overflow? And then Josh was like, hey, we actually have him coming on the show next week. And I said, oh, that's good timing. What is going on at Stack Overflow right now?

(Jody Bailey at 00:00:55) Yeah. Well, there's a lot going on at Stack Overflow, and the common conversation and probably the one you had is, to your point, what's going to happen to Stack Overflow? Right? And that is primarily because there's been a significant drop in traffic to the site. When you think about AI and what it's able to do for you, it's quick answers, no judgment, and oftentimes detailed things. But all the easy questions—quote, unquote easy questions—right, that people come to Stack Overflow for.

(Jody Bailey at 00:01:35) And oftentimes, as a developer, you would not memorize things, and you just go to Stack Overflow and just copy it because it was easier than filling your brain with it. All those things are available to you at your fingertips through any number of the different AI tools. So there's that. And then even over time, as it's gotten more and more difficult to ask questions on Stack Overflow that are unique, that stand the test of time, that meet all the criteria, people don't always like to be told that their question doesn't meet the standard. And as a community, we're not always friendly about the way we do that.

(Jody Bailey at 00:02:18) And so the confluence of being able to get quick and easy answers and no judgment and it being easier, friendlier, I think has contributed to the decline in traffic. Right? So what we're focused on now is, how do we address that? How do we bridge the gap? Where do we fit? How do we continue to support technologists and developers? And there's all sorts of ways of going about that. So our focus really is addressing exactly that issue that you're talking about. And a big part of that is talking frankly to people like you or other users and getting different perspectives. So when you ask me questions, I may ask you some questions in terms of things that you'd like to see because, literally, for me, my CEO, all of the folks—I mean, we're constantly talking to people and trying to understand the role that we can continue to play in the development of technologists, whether it's early stage career, late stage career, helping develop the skills, learn, be more effective, be more efficient, grow their careers, all of those things.

(Jody Bailey at 00:03:36) The other thing that I think we fundamentally believe is that there's a need for some kind of human connection. Right? Knowing that you're able to interact with other people and not strictly be isolated and interacting with technology.

(Joel Beasley at 00:03:57) I don't mind the dealing with the robots. So there's—let's talk about the use cases of it. So my experience is not everybody's experience. Mine's very narrow. Mine's I had a query. I would Google it or, you know, Stack Overflow would either have the answer or I'd post the question on Stack Overflow.

(Joel Beasley at 00:04:11) Get made fun of and then get an answer from somebody who was. Right? That's the trajectory. And now I just do it right inside of Cursor or whatever IDE that I'm using because the thing just helps me. I haven't had much of a need to go talk to a lot of other people other than asking friends that I already know, engineers that I already know, how are they solving it? It's like the social credibility thing. I know if I've got a question about this, I call Derek. I know if I have a question about that, I call John.

(Joel Beasley at 00:04:50) I know the different people and where they've spent their careers, and I'll go to them that way. So what are you hearing about? When you're talking to other people, what are they saying that, yeah, the reason why you would come to Stack Overflow versus just ask your in-editor LLM. Are they saying anything specific?

(Jody Bailey at 00:05:16) Yeah. So I think your use case is actually pretty spot on, right, in that most people are getting most of their answers through the chatbots, AI, and it varies based on where you work and the adoption and all that. But I think perhaps a unique advantage you have, especially having been at it for a while, is you have that network of people. You have your Derek, your Dave, John, whoever it is to go and ask. Right?

(Jody Bailey at 00:05:47) But if you imagine somebody that's newer in their career or doesn't have that network, who are those people that they ask? Right? Obviously, the first thing you do is you try to ask people that you work with, but it could be that they don't have the expertise, et cetera. And so we believe that having that type of network of people that's outside of your own network is really valuable. And one of the premises that we're exploring and I believe pretty strongly in is, when you think about that network of people, establishing your community, and a million people is not a community.

(Jody Bailey at 00:06:28) Right? And so if you think about Stack Overflow as a whole, most people don't really feel a connection to the people there. Right? I mean, we have different sub-communities, especially what we refer to as Meta. So it's the moderators and the folks that are super active on the site.

(Jody Bailey at 00:06:47) So one of the things that I think we need to do that we're looking at is how do we create smaller communities, sub-communities, where you can have more experts and have a more personal relationship? And how do we do that in a less intimidating way? Right? So do we allow you to post a question that maybe doesn't meet the standards of the core canonical Q&A? Right?

(Jody Bailey at 00:07:17) Like, create a space that's separate from that that allows you to have more casual conversations, to ask questions that maybe have been answered before, or maybe it is a homework question. Things that don't traditionally belong in Q&A on Stack Overflow, creating a space for those kinds of conversations and establishing a network of peers or at least identifying the different areas of community that you can get help in different topics that maybe you don't have the expertise in.

(Joel Beasley at 00:07:49) And you guys are a lot more than just coding because there's Stack Overflows for builders, engineers, mathematicians. Right? There's all sorts of ones. Right?

(Jody Bailey at 00:07:59) Exactly. Yeah. We've got, I think, 187 different—we refer to them as Stack Exchange sites—but a number of different communities. And it's everything from technology to cooking to religion, et cetera. So there's a number of them, but there are literally dozens of different technology sites beyond stackoverflow.com.

(Joel Beasley at 00:08:23) And then you have it for Teams too, like internal products where I can basically have Stack Overflow, but at my company with my people.

(Jody Bailey at 00:08:31) Exactly. So Stack Overflow for Teams or Stack Internal is exactly that. It's the ability to create your own version of Stack Overflow within your business or enterprise. And it's a very similar format, same idea of being able to ask a question and get it answered. Clearly, you know, the—

(Joel Beasley at 00:08:52) That might be an interesting way to train LLMs. Right? That might be an interesting value prop. Are you seeing anything like that? Like, hey, put this in your Teams, have them answer all your questions, and that's going to be used to train your LLM for the future.

(Jody Bailey at 00:09:02) Yeah. Exactly. So when you think about using Cursor or any of the other AI tools within an organization, many of them are using the LLMs that, candidly, we're providing content to. Right? So we've shared—OpenAI leverages Stack Overflow data for their LLMs. Right? And we support most of the large LLM companies. So a lot of the content from Stack Overflow is already available from us through those means. But when you're working within a business or an enterprise, you often have your own standards, your own ways of doing things, and those don't exist in the models. So how do you factor those things in?

(Jody Bailey at 00:09:47) And what we think of it is kind of a knowledge intelligence layer, if you will, where we can still use the community mechanisms, but strictly within an organization. Right? And then we can either make that data available if they want to do fine tuning or build their own models, or we're in the process of releasing an MCP server for the internal product where we can ingest content, use the mechanisms from all the different places that you store information and then surface that out to the tools via MCP. So we think that's a really big opportunity in terms of helping bridge the gap between publicly available knowledge and private knowledge.

(Joel Beasley at 00:10:33) And for the people who don't know what MCP is, can you just give a brief overview of that?

(Jody Bailey at 00:10:38) Yeah. So Model Context Protocol, the idea of it is establishing a standard that lets different AI agents and AI tools interact really seamlessly. Right? It's probably a poor analogy, but I kind of describe it as like HTML for web pages. You have a standard, and it makes it really easy to interact.

(Jody Bailey at 00:11:01) You might have a better description, but that seems to work when I'm talking at least to the non-technical folks. This audience, it might be too generic.

(Joel Beasley at 00:11:11) No, no, I think that's good. That's good. Yeah. I—look. The other thing too is as I've been around my career long enough to see how new things emerge, everybody's trying to grab for phrases and trying to find the right, what do we call this thing? And then you watch, there's usually the top two that pan out after a year or two, and then one usually becomes dominant. And so there's like a good one to five year life cycle of the new technology before everybody's cool with the exact words we use to describe it. It just takes time.

(Jody Bailey at 00:11:45) Yeah. And I mean, when we first started talking about MCP, which, yeah, seems like forever, but it's based on the new version of time that we're talking about with AI innovation. You know, it was, what, six, eight months ago or something, and I had the same kind of thought that maybe you're expressing similarly. Like, there are so many standards that kind of come, and then they're not standard.

(Jody Bailey at 00:12:16) Right? It's like the classic, you know, VHS versus Betamax or whatever going way back. But so at first, I was a little hesitant to jump on the MCP bandwagon, but it just—I mean, it seems like it's taken off, and it's really proliferated.

(Joel Beasley at 00:12:42) Oh, for sure. Seems super straightforward. You know, we've got a version that you can easily add to Cursor or Copilot, and it's super easy to add and interact with and get information from Stack Overflow directly.

(Joel Beasley at 00:12:56) Nice. Yeah. I mean, AWS has MCP servers, GitHub, Oracle. They're all doing it. It's becoming popular, and I'm sure it means different things inside of different communities too.

(Joel Beasley at 00:13:05) Because I heard some of the blockchain community talking about MCP stuff, and it's kind of—but it's a little bit different. And so it's really specific. But the way that I understand it is it's just like this bridge between AI models and external resources. So it's the glue that's going to allow AI agents to do stuff with your API or your database or whatever it is, but just this bridge between them. And everybody's going to pile on, build all these MCP things and connections, and you're already seeing it with the Zapiers of the world and the make.coms of the world. And they're all pushing forward hard on this.

(Jody Bailey at 00:13:45) Totally.

(Joel Beasley at 00:13:47) So you guys see AI as collaborative or competitive, or how do you see it?

(Jody Bailey at 00:13:53) Yes. Yes, sir. Yeah. It's a little bit of both. Right?

(Jody Bailey at 00:14:00) So it created a really unique financial opportunity for the company in terms of being able to license access to the content to the large language models. Right?

(Joel Beasley at 00:14:13) Oh, you guys license the content. Oh, okay.

(Jody Bailey at 00:14:17) Yeah. Exactly. Nice. So that's why I say, you know, they're using our content. I mean, everybody—pretty much all the big players are using it. Most of them have gone through and are licensing it. So on the one hand, it's been fruitful and beneficial that way. On the other hand, we talked about traffic going down and the need for Q&A in that format declining. So there's that aspect, but then there's also, how can we leverage it to help serve the people that are looking for help more effectively and efficiently? How can we, like a lot of businesses, use it within our own teams and organizations?

(Jody Bailey at 00:15:06) So my teams are also leveraging Copilot and Cursor and such. So we're using it. We're looking at it as a competitor because it does take traffic from our site, but then that also creates business opportunities. So it's kind of this weird mix-up of things in terms of how that all works. So I think we're exploring different ways of leveraging it and using it.

(Jody Bailey at 00:15:37) On stackoverflow.com, the moderators, the people that run the site have said no AI on stackoverflow.com. Right? And so we have had different experiments and run things there. But in terms of content, that's like a non-starter. We don't want AI-generated content on the site.

(Joel Beasley at 00:16:05) Oh, so AI can't respond to my Stack Overflow question?

(Jody Bailey at 00:16:08) So what we're doing—wait.

(Joel Beasley at 00:16:14) This is where it gets fun.

(Jody Bailey at 00:16:15) Alright. You can't—

(Joel Beasley at 00:16:15) Let's do it.

(Jody Bailey at 00:16:16) You can't post an answer generated by AI. Right? So—yep. And let me rephrase that. You'd be in violation of policy if you posted an AI-generated answer.

(Joel Beasley at 00:16:32) Okay. See? Now we're talking. Okay.

(Jody Bailey at 00:16:35) Right. So as the organization responsible for providing the platform, we don't want to create tools that encourage you to generate an answer. Right? We want answers to be generated by humans.

(Joel Beasley at 00:16:51) I gotta get those points, though, dude.

(Jody Bailey at 00:16:53) Exactly. Great points.

(Joel Beasley at 00:16:57) I remember when I was really early in my career, Jody, I knew just enough to help the noobs, the people that were just newer than me on Stack Overflow. I was 10% on a scale of zero to 100. So I knew just enough to help people that are just trying to get started. I would always be the nice guy. I'd always be like, hey, you know? And try to be that person. Try to get those points up. You know? So frustrating when I answered it, but someone smarter than me answered it better, and then they get the points.

(Jody Bailey at 00:17:28) Well, then there's always the fastest gun in the West. You know? Like, who can get the first answer?

(Joel Beasley at 00:17:35) Yeah. Yeah. Oh, yeah. Yep. That's before kids. That's before wife and kids. Before you—

(Jody Bailey at 00:17:44) Before you had a well-rounded life.

(Joel Beasley at 00:17:47) Yeah. The Stack me sitting on Stack Overflow now is like—there is a group of people who are currently at a stage in life where they can contribute, and that's great, and it's helping move things forward. But right now, I'm outside of that stage.

(Jody Bailey at 00:18:03) Yeah. Yeah. One of the things that we're exploring too is the whole reputation system incites a lot of motivation for a lot of people in terms of contributing. At the same time, you can have somebody that answers one question that's wildly popular, and it could be from ten years ago, and they just keep accumulating reputation.

(Jody Bailey at 00:18:30) One of the things that we'd like to explore is how do we make that reputation mean something besides you had one great answer? How can we use that to help you demonstrate your expertise, your proficiency? And how do we do that over time? Because one of the things that we used to have was the Stack Overflow job board, right?

(Jody Bailey at 00:18:55) And it was a way of showing your work. You could—it was really geared towards developers and demonstrating real proficiency as opposed to an endorsement on LinkedIn, for example. Right? And so we'd like to reevaluate that and see is there a way that we can adjust reputation to support that notion of actually representing proficiency and skill and not just you happen to answer one question ten years ago that now gives you 30,000 points in rep. So we believe that if we can do that, then that's another way that we could support technologists, especially as they're developing new skills.

(Jody Bailey at 00:19:39) If we can provide another way of demonstrating proficiency, then hopefully we can help them find new jobs, find new opportunities, et cetera.

(Joel Beasley at 00:19:51) What do the people want? What are they telling you that they want?

(Jody Bailey at 00:19:55) You know, the thing that people are telling us that we want is tricky. Right? And I should qualify this—we're talking to lots of different people, and there's lots of different personas. If you talk to our meta group, the folks that are managing the site for all intents and purposes, they're like, don't break it. Just clean things up and make the existing better.

(Jody Bailey at 00:20:18) But the thing that I hear most often when talking to developers that are using AI is, hey, look, I'm using the AI and sometimes I'm not sure if I'm better off having it generate the code for me and then editing it or just writing it from scratch.

(Jody Bailey at 00:20:44) Because oftentimes I generate all this code and I get to a point where I'm kind of stuck, and it's all this code generated and I don't necessarily—and there's so much of it—I don't necessarily know what's going on. And how do I fix it or figure it out? So the thing that people ask for is, hey, can you help us with that? And that's tricky.

(Jody Bailey at 00:21:05) Right? It's like interpreting the code, interpreting the intentions, and then trying to provide guidance on what it's doing, what it means, and how they should approach it.

(Joel Beasley at 00:21:16) Well, and that's largely going to come from—and you can correct me if we're not talking about the same thing—but that was my first experience too. When I had ignored it for quite a while, and then because of the show and talking with people and watching it, I said, fine, okay. I was like, I'm going to do a weekend project with this new tech. This is about a year and a half, two years ago.

(Jody Bailey at 00:21:34) Yeah.

(Joel Beasley at 00:21:34) And it was horrible, and it was horrible. I was like, this is—I don't get it. And then I'm like, okay, well, if I don't get it, I mean, I'm sure other people don't get it. Let's go see what people are doing. I started to watch some YouTube videos, and I was like, oh my goodness, I was using it wrong. Right? And then you watch people who were really good at using it to build stuff, and you're like, oh, okay. You do it in these small little chunks, and you do it like this, and you set up your environment like that.

(Joel Beasley at 00:22:00) And then you don't get into that point where you're super far down into the—you're just programming differently. And once I figured that out, I was like, this is very powerful. And most people will dismiss it because you have to get past that first layer of learning how to use this tool.

(Jody Bailey at 00:22:20) Yeah. That makes a lot of sense. The other thing—and tell me if I'm interpreting this right—but the other thing that I believe to be true there is that you have to understand how to break things down and actually implement it at some level. Right?

(Jody Bailey at 00:22:39) Because it's really easy to describe a requirement and have it generate something for you. But what I'm hearing you say is that you're thinking about that requirement and what are the different pieces that need to be put together in order to actually fulfill that requirement and then generating those components—loosely used. Right? But breaking it down into smaller pieces and implementing them. Is that right?

(Joel Beasley at 00:23:06) Yeah. Well, I mean, yeah. It's also like, what level of person are we talking about? Because I could see how newer engineers would be very confused if all this code's getting generated. And I've been doing this for twenty years. Like, I have a very clear understanding of how this stuff works in the context of my stack. I don't know all the stacks. I don't know all the frameworks, but I know a couple of them to do the things I need to do. And so to do what I already know how to do, but use AI to do it, is like, just supercharges me. It makes me like a team of ten. It's unbelievable.

(Joel Beasley at 00:23:36) But I could see how newer engineers would be like getting lost. Like, I'm not lost at all when I'm having it generate stuff for me. I'm checking that it—my job is checking that it did what I wanted it to do correctly.

(Jody Bailey at 00:23:48) Yeah. And that's exactly what I'm getting at, is because you have the experience, because you've built things, because you understand, you're able to kind of break it down into those smaller pieces where somebody with less experience may not have the knowledge or experience to know how to break it down, and that's where they get into trouble. And that's kind of the question that comes up all the time. Right? It's like, well, if AI is doing all this coding and it empowers junior developers, how are they going to become senior developers?

(Jody Bailey at 00:24:21) Right? And I think that's one of the big challenges that a lot of folks are seeing, especially when they think about, okay, well, everybody can write code. It's like, yeah. But it gets tricky when you generate all this code and you're trying to figure out how to debug it or fix it.

(Jody Bailey at 00:24:39) Right? Like, a friend of mine, he's CTO at another company, started off building games for Microsoft. And so he decided on a weekend to build a game. And yeah, he got it built, but then there was like this one issue. And he spent hours trying to use prompts and stuff to get it to fix.

(Jody Bailey at 00:24:59) And it was something that took him ten minutes looking at the code, understanding it, and fixing it. And I think that's the tricky part is how do we ensure that we're educating newer developers to understand those pieces in order to be able to decipher, debug, and actually get what they want?

(Joel Beasley at 00:25:23) Hey, until then, they're just creating a bunch more opportunity for me, so I'm happy. I'm happy for that. Right? Yeah. Go build the next billion-dollar product and not understand how it works at all. You know? Yeah. That's great for me. Like, congratulations, you know?

(Joel Beasley at 00:25:36) Yeah. But that is a trade-off people are going to have to make too. I get a lot of questions, Jody, from people with their kids in high school and stuff. And, you know, my neighborhood and social circles, like, what do I—what do they do? What should they learn?

(Joel Beasley at 00:25:49) And I'm like, I don't know. Like, I don't know what to tell you. Yeah. You know, learn—and so what I've come back to is like, learn—understand value creation in a marketplace and understand how currency works because that's how we transmit value creation and that's the economic system. And if you understand those two things, then you at least need to understand those. Don't worry about like, should I learn Ruby or should I learn prompting? Like, understand value creation and currency because those are the two most important things.

(Jody Bailey at 00:26:23) Interesting. Yeah. I love that thought process. I hadn't really thought of it that way. Like, how do you envision people doing that? Like, what does that mean exactly?

(Joel Beasley at 00:26:36) Well, I would start with just a history class on currency. Just Google currency 101 history class or something, you know, and or go to the currency Stack Overflow exchange history.

(Jody Bailey at 00:26:49) Do a little—

(Joel Beasley at 00:26:51) I'm not going to say ask your LLM because I'm going to be respectful right now. No. Okay. But yeah, understanding the history of currency and why we have it and when it was created and why it was created and what the benefits are. That'll teach you about currency, which is the network in which we exchange our value and value creation itself. I don't have a specific resource off the top of my head, but I dove into it in my early twenties when I wasn't getting the results that I wanted out of life.

(Joel Beasley at 00:27:23) And then I realized, like, well, I thought I was very valuable, and I am to my parents and my friends. I am not economically very valuable. Yeah. So there's like the human, my soul, like my spirit, like who we are as people, Jody. We're like on the same level. We're both valuable as a human life and as a person, but then there's economic value that happens in the economic system. And the cool thing about that is you can actually choose to become more economically valuable.

(Jody Bailey at 00:27:53) Right.

(Joel Beasley at 00:27:53) But you have to understand like the game that's being played. Yeah.

(Jody Bailey at 00:27:57) Interesting. Yeah. It makes me think of the book Rich Dad Poor Dad. I don't know if you've ever read it. Yeah.

(Joel Beasley at 00:28:06) We have that book. We also have the board game for our kids. Have you—did you ever see that?

(Jody Bailey at 00:28:11) No. No. I fear that I fall more in the poor dad category, right, where I just work harder.

(Joel Beasley at 00:28:21) That'll happen. Yeah. Yep.

(Jody Bailey at 00:28:25) Yeah. Then you hope you're in the right place at the right time providing the right value. Right? And as they say, hope is not a strategy.

(Joel Beasley at 00:28:33) That is true. And I was doing an interview—like, I was on someone else's show, and they said when they asked me that question, they're like, just don't give me the answer, learn to prompt. And oddly enough, that was my answer I was going to give. I was going to be like, look, the thing you need to learn is you need to learn how to prompt because that's an incredibly valuable skill today.

(Joel Beasley at 00:28:55) But I realized in the past year and a half that these machines are becoming so much better at making it so you don't have to be good at prompting. Right. Understanding your intent, figuring out the prompts and the systems that need—like, Grok is doing a great job of it. Have you been playing with that?

(Jody Bailey at 00:29:13) Just a little. Not a ton. But that's for me is one of the big questions, right, is obviously the rise of generative AI has really gone fast, and the quality has improved tremendously. And the question is, if we're 80% there in terms of being able to get you all the things you need to write the code and prompting and sorting that out, can—is the last 20% the 80% of the work? Right?

(Jody Bailey at 00:29:50) Like, how close can we actually get? Right? And I'm not sure if I'm making sense, but you're approaching 100%, but like, is it ever going to get there? How close can it get? And is it going to happen in the next three years? Five? Fifteen? You know, I don't know. But that's one of the things that I think about all the time in terms of especially our strategy in terms of that knowledge intelligence layer and the necessity for humans. It's like, I believe that.

(Jody Bailey at 00:30:27) I don't think that we're going to get all the way there in the next five years. But, you know—

(Joel Beasley at 00:30:33) Well, I mean, look at humans and machines like lifting things, and we've got Caterpillars now and like all the big heavy equipment. Like, we don't hire hoards of men to lift these things. We rent a catapan, you know? So that—yeah.

(Joel Beasley at 00:30:49) It's changed, but man still has jobs. Like, there's still—there's new things to be done. One thing that gives me hope, because I am an optimistic person, Jody, is that while I believe there will be a massive consolidation in the tech sector and knowledge work as a whole—this goes to lawyers and doctors, and there should be massive consolidation, like, meaning one really smart physician or engineer with the AI being able to do the work of many more. Well, that's true.

(Joel Beasley at 00:31:21) And that may start to get a little nihilistic, and that's the path I went on. I was like, okay. Well, that's really bad. And then, through prayer and other stuff, I was like, okay. You know what? Something I'm not thinking about is with these tools being better, we will have large breakthroughs in new sectors creating new jobs. So think about small modular nuclear reactors becoming a thing and cleaner energy, opening up an entire new supply chain and infrastructure for everything from shipping the nuclear reactors to designing them to the material that goes in them to all of that. That'll be a whole new stuff. We'll get new advancements in biology with new medicines, and that'll create whole new cycles and distribution chains. And there will be all—I think this intelligence will help us create and have breakthroughs in so many areas that there's still for my lifetime, there's still going to be more than enough work for people.

(Jody Bailey at 00:32:16) Yeah. I tend to agree with that, and, you know, I'm also optimistic. I also probably have fewer years left, so, you know, my forecast doesn't—

(Joel Beasley at 00:32:28) Do you have kids, though? You've got kids?

(Jody Bailey at 00:32:30) Yeah. Exactly. Yeah. But I agree. Right? I mean, I think every generation, every transformation, there's always this concern that, oh, everybody's going to be put out of work. And historically—and, you know, almost everything seems to repeat. Right? Historically, there's always been the generation of new work, new opportunities, and we still haven't reached a point in time where everybody can just kick back and just exchange goods or get some sort of social income to live on. Right?

(Jody Bailey at 00:33:12) And I don't see that happening anytime soon. I don't think—

(Joel Beasley at 00:33:17) That's very healthy for humans. You made a lot of money in your life.

(Jody Bailey at 00:33:22) No? Oh, I'm not sure where you're going with that. Yeah. But yeah.

(Joel Beasley at 00:33:27) Oh, well, because like, when you make a lot of money and you get to the point where like work is optional, that's a very scary place to be. Yeah. You have to figure out systems and habits and like, you have to really—I mean, yeah. Do you disagree?

(Jody Bailey at 00:33:42) No. I agree 100%. And—

(Joel Beasley at 00:33:45) Oh, okay.

(Jody Bailey at 00:33:47) I'm one of those people that doesn't do well without some sort of goal or purpose or something that requires effort. Right?

(Joel Beasley at 00:33:57) So—absolutely. Yeah. I don't—the dream for me is not being on the beach doing nothing. The dream for me is achieving a difficult dream.

(Jody Bailey at 00:34:06) Yeah. Yeah. Well, it's funny because one of the questions is, okay, well, you know, Stack is going through this existential crisis, if you will.

(Jody Bailey at 00:34:24) Why would you want to be there? Right. And it's kind of that, right? It's like, I mean, it's hard. It's a challenge.

(Jody Bailey at 00:34:36) It's for a good cause, or I believe it's for a good cause. And those are kind of the key things for something to me that's fun. Right? It's like I want to do something that's hard, that's meaningful, and that is ultimately rewarding and can lift other people up.

(Joel Beasley at 00:35:01) I feel like, if have you crowdsourced ideas?

(Jody Bailey at 00:35:07) I mean, yes, I guess. Right? Like, we do a lot of surveys and feedback from our community members. We talk to customers. You know, all of us are pretty much talking to people all the time. So we are doing that. Do you have another mechanism you're thinking of?

(Joel Beasley at 00:35:33) No. I was just curious. Like, I have found that when I talk to my group, I'll start to hear one thing emerge, and it'll be very clear. One of the questions I had that would help me understand better in this conversation is where does 80% of the revenue come from for Stack? Is it advertisers? Like, people advertising on the site, or is it Teams product? Where is the money coming from?

(Jody Bailey at 00:35:59) Yeah, it's distributed. So ad is actually the smallest part of our revenue. Then the next two biggest, at this point, I'll say they're roughly equal. It's the data licensing and it's the Stack Overflow for Teams. So we've got three different revenue streams. Advertising as well as employer branding is another, which is kind of just a different type of advertising. Historically, over the past five to seven years, pre-AI, everybody in terms of company valuation was thinking software as a service. And the public site really, prior to AI, the only revenue really was advertising. We had a job board, but that didn't seem to have opportunity for a lot of growth or ongoing value from an evaluation of the company perspective. And so we spent a fair amount of time investing in the SaaS products, so Stack Overflow for Teams. And the AI change came that impacted both sides of the equation. And so we're reevaluating that. One, I think we underinvested in the public platform to some degree in terms of making sure that it kept up to date with what people needed and just sometimes basic blocking and tackling. And then we got a little spread thin, and the AI kind of disrupted or is disrupting not just public, but also the enterprise solutions. And that's why I talked about this new path that I'm actually really excited about where we're able to help validate the private knowledge, leveraging the internal communities and then making that available more easily, right, or more readily. Trying to meet the developers where they are. Right? So whether that's in Slack, whether that's in their IDE, really trying to make sure that they can get access to the information regardless of where they are.

(Joel Beasley at 00:38:37) Interesting. I think that Stack for Teams, private knowledge to be then used to train or fine-tune LLMs, I think that is a win because you guys pioneered that data collection process. It's a huge strength that you have and that interaction, and companies will definitely need that. So it's good to see that that's growing. Are they, they have, they pay for the Stack for Teams, but does that already include that model stuff that we just discussed, or is that something they pay extra for? Or is it coming soon?

(Jody Bailey at 00:39:17) Yeah. Some of this is in process. Right? Because one of the challenges any business or enterprise I've been a part of is the information and knowledge is kind of spread all over the place. Right? Whether it's Jira tickets, whether it's a Confluence page, whether it's a Google doc, whether it's in the code. And how do you, besides just question and answers, which is what we've done historically. Right? Is, okay, rather than asking Ash, how do I do X, Y, and Z on Stack Overflow for Teams? How do they get that information? Right? Right now, they might look for it in all these other places and then ask somebody or vice versa. So the thing that we're doing now, and we're just piloting this with a few of our customers, is to actually be able to ingest content from all these different places and then reformulate it into questions and answers or other types of content, identify subject matter experts across the organization, and then determine based on a number of different factors as we're importing it. How confident are we that this is accurate? Is it current? Is it the only source? Was it authored by an SME? Okay. Probably a high confidence score. We can ingest that and probably don't need to review it. But if it falls below a certain threshold, then we want to have people review it and say, yes, good, not good, whatever. Right? And I'm oversimplifying things, but the idea is how do we make all that content available in a way that's easy to say, yes, this is right or wrong, and then surface it up to folks when they need it, where they need it so they don't have to figure out where to go look for it or if it's the right version, if it's up to date, et cetera.

(Joel Beasley at 00:41:22) I think that'll work. I think that's how you guys save Stack Overflow.

(Jody Bailey at 00:41:27) I'm excited about that. I think that's our revenue model. And then on the public side, figuring out how to support people in their own journeys. Right? And I think there's always gonna be—

(Joel Beasley at 00:41:45) But you have to, it's like you gotta do the life raft thing. You gotta, like, save yourself first. That's, like, step one. Make sure you're still alive. Yes. I'm not shy about that at all. Yeah.

(Jody Bailey at 00:41:59) I love that. I'm gonna borrow that. Thank you.

(Joel Beasley at 00:42:04) Yeah. I was, I stole it as well. But, yeah, that's so important because if you guys, and by the way, I think you'll do well with that, what you described, because it fits really easy in my mind. Like, and you also have the brand trust of Stack Overflow. So when you start to go to market with this in a deeper way, you get past a lot of startup, you're not a startup. You're just not. Right? Which is actually a huge strength. Right? You're Stack Overflow. You're known for being intelligent with this stuff. And so, I think you'll, how's your, how many people do you have at Stack Overflow?

(Jody Bailey at 00:42:40) It's about 350.

(Joel Beasley at 00:42:44) How many of those are sales?

(Jody Bailey at 00:42:46) Probably around a hundred. I mean, when you factor in different ads, different, and I could be off. It's been evolving. But that's, customer success, support, sales. We've got three different products that we sell: advertising, data licensing, and then the Stack internal or Teams solution.

(Joel Beasley at 00:43:18) Nice. Yeah. I'd say of those, the Stack for Teams, I think, is gonna be the winner. Ads is scary because it's dependent on traffic, and it's already a lower point of revenue. Data licensing sounds great, but there's gonna come a time when synthetic data means they're not gonna really need you. But Stack for Teams, that sounds like, okay, they're gonna need this, and you're gonna get into their deployment pipeline for lack of a better term. And then they're just gonna be like, that solution, that thing's been solved, and we just pay Stack for that solution.

(Jody Bailey at 00:43:46) Exactly. Yeah. And that, I mean, that's, yeah, kind of what we've been trying to figure out for years. Right? It's like, how do you become part of the pipeline? How do you become an essential product and not just a nice to have? Right? And we've always been a nice to have. Right? It's like, okay, well, yeah, the developers like it. We believe that it improves efficiency. But if it comes time to cut budgets, it's like, well, do we really need this tool? And I think you're right. If when we're able to provide that knowledge intelligence layer and when we get embedded in the workflow as a natural part of doing people's jobs, we become much more valuable to the user, the developer, as well as the company. And then to your point, it's, like, much less likely to be just disrupted or removed when budgets get tight.

(Joel Beasley at 00:44:47) Yeah. No. I, that's awesome. Okay. I have hope. We may rename the episode. There's hope for Stack Overflow. Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.