[0:00] What is Stack OverflowHello, the guest today is Prashanth Chandra Segar. He is a Masters of Engineering Management at Cornell and he is now the CEO of Stack Overflow, which is a really big website and technology company. So hi Prashanth, what is Stack Overflow? Yeah, Thanks, Tony. Awesome to see you again. I was awesome to see you in our offices here when you came by a few months ago and always love your enthusiasm for the Cornellys, as you like to say. So, you know, go be great. Yeah. So what is Stack Overflow? Stack Overflow is the the world's largest software developer community and we've been around since 2008. But we exist basically to serve the world's developers and technologists so that they get access. Democratize that to the world's information and all things technology, programming languages and technical topics and so on. And we have something like 50 million questions and answers on every possible technology topic. And we have, you know, hundreds of thousands of questions being asked, you know, every week, month and year. So it is, is a very robust place where technologies from all around the world have come together to engage and it basically get unstuck when they're stuck either in when they're generating code in their AI application or their writing code and they need to sort of understand it deeper. We are the world's experts on JavaScript and all these sort of, you know, great web frameworks of private programming languages that, that are there to help each other out. And so that's the that's the foundation. And beyond that, we obviously have several businesses around that. Our biggest business is focused on Enterprise Oregon. The companies or what we call Enterprise Products, where we sell a private version of Stack Overflow into companies so that they can effectively get the same outcomes inside their companies to share and collaborate with the experts within the company. All the sort of the critical knowledge that they need to sort of move their organization faster, especially in this AI world. So it's used as an amazing a platform to drive even adoption of AI tools. How do you use the latest of technology inside com And it also is used as a very sort of accurate knowledge base for AI assistance and Co pilots. And soon AI agents had to start using for, you know, to to really sort of do great things inside the companies to drive up productivity and efficiency. So that's the primary business along with, you know, we we license our data. That's a secondary business. And then we we also have advertising, which is the function of us being a large content site. So that's that's us in a nutshell. Like to people who don't know type overflow, what's the example of a question? What's the example of the answer that? Yeah, on the public, on the public platform, the one that has 50 million questions and answers, it's every possible nook and cranny of, you know, you're trying to write code. So, for example, you may be trying to figure out just the basics of JavaScript. You may be a very early programmer and JavaScript, you know, the world's most popular web framework and people, it's been around for a long time, but we have literally hundreds of thousands of questions on every possible dimension of unpacking that right? So if you're stuck even when you're getting started, how do you just get started with with JavaScript as an example? But there's that basic sort of premise all the way to some very detailed questions on errors that people get when they're writing code, let's say in an IDE or what is known as integrated development environment for the writing code or generating code these days in an AIIDE, they're invariably going to have encounter an issue of some kind. There's going to be an error. There's going to be a syntax error where you didn't close the parentheses, or you know, you got the format wrong, or you, you know, just didn't like correctly write the function out or didn't invoke the right library, whatever it may be, it is going to reference. There's going to be all sorts of details on those sort of subjects on the platform so that people can debug and get unstuck from what they're doing. Usually answers the questions. Are they also developers too? Yeah. Again, I'll take it from the public and the private version of our company, because again, the private version is our enterprise product. So on the public side, it's the world's developers technology. So we have a core group of very ardent users and contributors and the world's experts on all these technology topics and programming language, etcetera to spend a lot of time making sure, because we're all joined with the same mission of it's all about cultivating community, powering learning and unlocking growth. So we're all united in that admission, our community of 100 million people and our company and our employees. And what we try to do is make sure that the platform represents that. You know, we're really trying to connect experts with other experts in the spirit of unlocking people's learning by documenting the world's information on all things technology topics. So people are asking who have questions when they're in their flow of work, and people are answering who have the knowledge to be able to help other people. And they're all part of the same community. Some are more experts and some are more beginners, but you also see the flip where people are answering, you know, each other's questions in various dimensions. I've always been impressed with how hopeful people are on the public version of Stack over Full. Well, like, what incentivizes these experts to give their time to answer these questions from the beginners? Yeah. I think it, you know, it's, it's it's quite fascinating that it is ultimately in the spirit of moving humanity forward. I mean, that's basically what people just don't. They just believe that we all should be, you know, helping each other out. And anybody was written code or tried to sort of debug it on their own, knows how frustrating of an experience that can be, especially in the olden days. Things have gotten easier and easier over time, thankfully, you know, thanks to that Stack Overflow, when this company was founded by amazing founders, and even now with AI, which we power, all the LLMS now leverage our Stack Overflow data to produce that sort of that key aha moment where you get unstuck when you're actually in the flow of your work. So the user interface has changed over time, meaning previously we used books back in the day to solve the problem, then we used, you know, Stack Overflow, and now we increasingly use Stack Overflow, but also AI tools. So it's sort of a progression of how people are trying to get on steps. So the incentive is really to help each other out. And it's, I think it's somewhat unique to this community in that people genuinely want to democratize information about technology because it's a fun thing to go build and to prevent the frustration of having to do it all on your own and to leverage actually the collective community knowledge to be able to move forward. Is it like, you know, Wikipedia, they got like these groups of like power users who've always been like updating and they get like awards next to their name. Is there any type of incentive like that too? For sure, of course there are. I was mostly describing the intrinsic motivation, which is to help each other out. Of course there are extrinsic motivations. And so yes, we certainly recognize people for their expertise, for helping a lot of people out. Hey, the world's best JavaScript programmers are absolutely recognized or Python programmers or Ruby or whatever the topic is, or us, which is, you know, even more popular these days. All these languages and frameworks, etcetera, are people are recognized for the contributions to the other fellow community members and the impact of their contributions. So for example, if you, you know the answer that you wrote maybe like a few months ago or years ago, and you can go see how many people that did that, did that actually impact, you know, we have you attracted Tony, you know, 100 million people because with your answer. And that itself is like fantastic. Because the more that happens, more people reference it. And now increasingly it's being referenced through AI because our content is showing up in, let's say, ChatGPT or in Google Gemini in the code assist functionality. And you can see the actual attribution back to the original users who contributed those answers. And every time that happens, people click on those and upload those answers. Those people are getting recognized as experts. And that, by the way, proves to their personal profile on Stack Overflow they can very proudly show to, you know, future employers. And of course, it's a sense of personal pride as you're constantly learning in your career and you're progressing in your career with, you know, kind of continuous learning and impact, of course, to your fellow community members. So yes, absolutely. There's intrinsic badges and rewards and point systems and all the things that you you would imagine would be exciting for people. So this is all for like a public. How does this translate to the private one? Are are like employees within companies who are doing their jobs? Are they just, are they similarly motivated to fill out those questions as well? Yes, excellent question.

[8:14] Private Stack Overflow internally for CompaniesAbsolutely. So in 2019, effectively the year that I came on board as a CEO, so it's been about six years since I joined the company. We, I came on board because there's an early experiment that kicked off right before I joined, which was that large companies like Microsoft and others reached out saying they would love access to a private internal version of Stackoflow because they love Stackoflow in the context of Outlook platform. Such companies started asking for it and the company decided, hey, this is an amazing way to really build an enterprise business to be really helpful. The same users have been using Stack Overflow, but they, we just help them in the context of their companies because it's again like this, you know, internal community that they can build with experts and knowledge. And so that's what we did. Now we have the world's largest banks and financial services institutions, the biggest tech companies, the biggest retail companies all around the world, you know, leveraging Stack Overflow as their private internal version, what we call Stack Overflow for or just to be soon to be renamed as Stack Overflow internal. That's what we're going to call it in a few few weeks. And so it's, it's an amazing platform to really have this human curated, expert, curated knowledge base inside your company, specifically mostly focused on technologists and tech adjacent groups. So it expands to developers, but also, you know, system administrators to product management or product marketing people, even sales engineering. Sometimes even companies use IT company wide, even in other departments and they use it as like an accurate knowledge base. And now we started noticing that a lot of these companies as they roll out AI initiatives inside their companies to introduce AI assistance and and so on to drive efficiency, we're seeing a lot of usage of our product being plugged into these AI assistance. So these AI, for example, you know Uber in the ridesharing company that I'm sure you know of, this is our product. It's a big customer of ours on Stack internal and they have built an agent called Uber Genie that effectively plugs into Stack Overflow internal, which has got thousands of questions inside Uber. They are able to then let this AI assistant go and answer a whole bunch of questions in their Slack instance internally. But it removes a bunch of, you know, bunch of inefficiency in the system where a whole bunch of real people don't need to go and answer those same questions over and over again. It's automatically addressing those, in some cases addressing internal support tickets that have been raised, and it automatically just clears them out, please them out with answers that are canonical, you know, correct ways to do things. So those are examples. But yeah, the incentive system is because people are very frustrated inside companies having to repeat themselves over and over again on the same topics. So there's always a need to find accurate information inside companies and also to be able to find it. And when you find it, it better be accurate, right? Like both of those things are big problems inside companies. And that's what our product does. It plugs into all their systems and it keeps things always accurate and trustworthy. That makes sense. I think in terms of like a company has a lot of information. The Slack has like a very specific type of like back and forth kind of information, back communication. Yeah. And then like, like, is there like the Wikipedia that they usually have as well? Like, do you see yourself as a wiki of information versus a back and forth kind of information? Really excellent observations. So systems like Slack and Microsoft Teams are like they're synchronous tools, right? Even this is a synchronous conversation, obviously, because you and I are speaking life. So we obviously have all used those tools and they're all this effectively, they're chat, they're real time, etcetera. But the asynchronous tools for collaboration are few and far between, right? There's the design space. For example, you didn't even think about a company like Figma that recently went public, like in the design space. But for us, this is a knowledge collaboration space. And so yes, what people have inside companies is typically for this sort of area. Typically they have internally built wikis or they have, or you know, there are certain wiki companies that are out there that you know, help them build internal knowledge bases. So those are the places that people currently have information. But our system is very unique because only does it have the ability to have canonical information with this Q&A pair and questions and answers, but also has the community curation and sort of the ability to keep information always high. So the content health is very, very high. The, the, the content will be always accurate and it's a content help to be very, very good. And so you can always trust it that the answer is coming from the the same authoritative and for the source. In addition, we are doing and we just announced this recently, a few weeks ago, is that we're able to now ingest information from other parts of the company, leverage this as a knowledge intelligence layer. So then you're using that human curation and expertise to make sure that you are effectively memorializing all the right stuff on our platform, and then you're able to leverage all that memorialized information anywhere in your flow of your work. What kind of informations are you reading in? Are you reading in GitHub or? Yeah, it could be. It could be get up information, could be Confluence, could be G Drive, Google Drive, it could be SharePoint, could be. We have a bunch of things that where people have information, they have files, they have other stuff, right. And that's used in very, very different ways inside the in our customers. A lot of our customers have technology related information on our platform. So similar to our public platform, which is always let's say programming language kind of web framework type of questions. These are more internal company knowledge. Could be, Tony, as simple as like how do I, you know, reset my password at company? Or it could be how do I spin up an AWS cloud environment within my company XP, right? So just the common questions you need to get started. There's a lot of different mechanisms in the product to get people up to speed very rapidly. So for example, we have something called collections. So a brand new person starts on a team inside a company immediately our product will send 10 questions and answers that are relevant for that, you know, let's call it developer one to be able to, to get up to speed very, very rapidly. So they're not spending weeks and months trying to figure out, you know, what to do with a lot of back and forth. But like really they're able to be fed with like here are the 10 things you need to learn like in the first week, you know, and then once you're off, you're off and running a lot faster. So your onboarding time is super fast. And so you want more, you know, your productive and hence innovation speeds increase and so on and so forth. Yeah, that makes sense. And the great thing is when people leave the company, the information still stays there, doesn't say in their head it stays here. Absolutely. So the company very important to see especially over the past several years, there's been so much you know, jobs will continue to evolve. There's going to be a lot of changes and there has been a lot of changes. There's going to be even more changes very soon in the technology area with developers and all all the knowledge that they are taking time to sort of share, etcetera. It doesn't just sort of like disappear into thin air, you know, once they leave the company, once they leave the company. So companies love the product because they're able to again, memorialize all the detailed thoughts that everybody collectively, you know, put together and, you know, the company keeps so moving forward despite changes that you may make if people change teams or anything else that happens. Yeah. Is the only way to use that by visiting a website? Or like what are the different ways that internal people use stack overflow internal? Amazing. Great, great question. It's well, you're speaking like a true engineer or physicist attorney. I like the nature of your question. So I like it.

[15:44] How to use Stack Overflow InternalSo you're right that our product, do you have multiple vantage points or multiple entry points? So you can either come to the internal platform which is Stack Overflow internal or you can access information in your Slack instance or you can access information in your Microsoft Teams instance or you can access the information in Microsoft Copilot or you can access it in GitHub. You can access it wherever you are actually. So we have effectively gone wherever the user is. So in Slack where, you know, you can say how do I reset my password? The answer is going to come straight from Stack Overflow internal because it's the authoritative answer that somebody in IT had provided, let's say a year ago. That's still accurate. And that's going to get fed to the new person asking that question in Slack versus that person have to come here or they can ask it in Microsoft Copilot. And we have done this integration with something called Microsoft Graph Connector where it's able to tap into all the knowledge and Stack Overflow internal and it serves it up when people are using Microsoft Copilot as one example in an AI context. So yes, we are heavily integrated our knowledge base and this kind of knowledge intelligence layer into all the different places that people spend time in inside companies. Interesting. GitHub might have their own data, and then teams might have their own data, whereas they're not merged together. So you merge it all together into one place. Correct. Yeah. And also it's very different kinds of data, right. Just to be clear, it's not overlapping those places because GitHub is like a code repository and and other things that they do along with that foundation. We are, you know, a context and knowledge place of human curated knowledge base, human curated context. And that is effectively, you need context and you need code. Code can be written, code can be generated with AI, but you still need a lot of context. So where are you going to have all, you know, what are the ways in which you can drive up efficiency by having the commonly asked questions on all, all things? Your code. That's one example. It's not limited to code within companies, to be clear. That's just one of the use cases. Yeah. Like cycle overflow, but traditionally has always been used by code, but now internally to be used for all kinds of Q&A questions. That is correct. If someone is like the CTO of a bank for example and they want to install this like the installation process like. Yeah, yeah, pretty pretty straightforward and actually you're right. So we do sell to CTOS or CIOs, but typically CTOS or VPS of engineering. And these folks want like any other SAS software, they are able to leverage it in a, we have a single tenant offering and we have a multi tenant offering and we have a very specific onboarding process where we get people access to the software. And then we obviously are able to invite users that they'd like to invite to the then knowledge is ingested, as I was describing with this knowledge in different layer. And then sort of the system goes and kicks off and we typically seed it with some amount of information. So the engine sort of kicks off like it's like kick starting your bike or starting your engine. And once you do that, then it's like off and running because then it's like, it's just like doing its thing, prompting people to answer like, hey, Tony, you're the expert on physics. You know, that's not a great example. I would say, hey, you're the expert on JavaScript. You know, Prashanth asked this question. Why don't you answer? It's going to do that all automatically. Be honest. Yeah. And by the initial ingesting of information, you solved the code start problem, because there's at least some answers to some questions initially. But yeah, you know, what's what's interesting about that is that so far what we have done prior to this knowledge ingestion work, people would we would work in a very sort of, you know, kind of close way with the customer to make sure that we to seed the knowledge base with some critical information. And it could be like, you know, companies have like 50, you know, 50 to 60 document repositories. You know, if you think we're like the largest companies, they've got some amazing amounts of historical knowledge that's been stored. And so they're all just like lying around and a lot of it is like just out of date and just collecting dust. So for them picking like a spot where we can say, OK, let's pick these, let's call it 100 documents and start with that for the most critical information. So we work with a few people on their side. That's what we're doing shortly. Now with this knowledge ingestion module that we just added, we're able to effectively ingest knowledge automatically and by saying, OK, let's click Confluence, GitHub, blah, blah, blah. And within that, let's make sure that we can give the admin the ability to say yes or no to certain documents. So the user it's not garbage in, garbage out. You can, you can exclude, for example, all the documents prior to 2020. Like, you know, it's not, oh, you have a lot of control, but it's a lot more automated. So we're removing the full start kind of concern that you're describing where there's not even human seeding anymore. It's like literally it's human verification to the automatic. Automatic seeding, so to speak. How do you do the like? Certain people are allowed to know certain information but not those people aren't allowed to know that information? Like how do you do the like the authorization part of this? Yeah, there's a lot of authentication. Firstly it has to do with user authentication components, but there's also the types of information that people are allowed to enter into this in research. There's a couple of admins within companies that say, OK, here's all the departments that get to use it and you know who's going to access where users, where they are specifically picked to be able to leverage it. Sometimes people want to open it up because they just want the community effect. They just want a lot of people to just sort of crossbow played and add and provide information so there's less restrictions typically on the one of two dimensions in terms of restrictions, the people admins decide, okay, only these out of our let's call it 100 departments. These 80 departments are going to get access to the system. Let's open it up to all 80 or it could be all 100 get access to it. Hey, but they're like, here are the rules on what you actually include here. You don't include X&Y, but you include you can all this by the way, we don't even see the data on our end. It's completely opaque to us. Like it has access to the data, but it's to your point, what do they want people to collaborate on and so that they don't share things across the company that they're not meant to. So for example, it's fairly rare for us to see, for example, legal departments are shared, I think, but we also have the ability to cordon off certain sub communities within the product that you buy. So that if you're, let's say, you know, we have certain customers that have projects that are quite confidential. So they basically carve out, you know, multiple sub communities where only those sub communities can engage with each other. And this is a very, very large company that we that we work. With so you have like each team can have their own little LLM agent. Correct. It could be extended to an own LLM agent because you could then sort of plug in your AI agent into that that repository to do various things. Yeah. And we also, by the way, we have integrated, we've built our own agent on top of Stackable Flow internal, which allows you to effectively it retrieves all the information from Stackable Flow inside your company, but also from the public Stack Overflow and provides it at the right place and right time. Either if you're in your ID, you're writing code, if you're on I'm looking up information, or if you are, you know, in Slack or in Microsoft Teams, all this stuff can show up. And that's called overflow AI. We launched that in 2023. This. Is a very big change from like

[22:35] Transformation from a B2C to B2B companythe Stack Overflow public that I've been used to back in the day. It's like a big change, like how did you bring the company into this new direction? Yeah, it's a good observation. Yeah, certainly. I think part of the mandate of me coming on board was to help build this business, which is the enterprise business. And the company obviously was prior to me joining heavily focused on advertising and talented job listings. So a lot of companies would post their jobs so, you know, they could find developers to apply to them. But we had to change the DNA of the company to be a lot more enterprise centric. So there was less of that in the company. And so it took a while. Over a couple of years. We brought on some amazing leaders, have some, you know, amazing executives that we were able to bring on. And in turn they brought on some other folks that and you know whether that is people who are approaching the CTO's to earlier point or serving customers in a customer success team that is you know fairly standardized in the SAS world etcetera, etcetera. So all the things you need for running a SAS company including demand Gen. and marketing and all these things you need to have those components in place. We had to build all that very heavily within the company. Some components of that existed prior to me joining, but they were all in the context of moving from what was talent and advertising to this enterprise motion. Because it's like a whole different team. You have to build up. You have to have these sales people exactly right. Well, knock on the doors now. Exactly right. Yeah. And so that is so literally we have two parts and 1 is what we call community products and one which we call Enterprise Products. So community products is what you're aware of, right. And I can talk a little bit more about what we're doing there as well. And then Enterprise Products is his entire enterprise motion, including, we haven't covered, for example, our data licensing partnerships with Open AI and Google and others, but which we stock also. So it's all in the enterprise side. As the two parts like diverge over time like, what's the common thing that holds it together? Yeah, it's a great question. So the our vision as a company is to be the most vital source for technologists. That's what we want to do, right. So there was a technology are on a public platform or inside the company. We just want to be vital, like trustworthy as in critical and you know, you can that's going to be very important in my mind and the context of AI because there's going to be so much AI slop and missing slop that's going to be floating around that you just don't know what's right, what's wrong. It's very obviously extremely impressive technology, but leveraging that along with the the expertise that is in our platform on a public platform as well as on inside companies is the best of both worlds. Human expertise comes in service of humans were combined with the AI functionality. So yeah, the link is that it's that vision that brings it together and it's ultimately trying to get people unstuck, right. While they're while they're trying to get stuff done. Yeah. That makes sense. And the good thing about technology is there's actually like a right and wrong, so there is like an actual curation

[25:30] Strategy for the Stack Overflow public communityprocess. That is correct. Now that's probably a good time for me to explain a little bit of the community product strategy that we have embarked for the past few years. And what we've done since, especially since the kind of the late 2022 time when ChatGPT came out, is that we started incorporating AI into stackable flow on the public. So a lot of debate with our community members whether we should include AI or not and which position should be banned. So what we've done is like not allowed for AI answers on the site because the hallucination issue that we just talked about, right? So that is we explicitly don't allow that because we want to keep the quality really, really high and that's important. There's been a lot of time and that's why it's been so valuable as a source and that will never change in Q&A. However, we have thoughtfully incorporated AI. So for example, in the question asking experience, you may have encountered this, Tony, when you asked your first question, ask any question, you know, you may have gotten slapped in the rest by somebody saying, look, it's a duplicate question or it's been asked already or something like that. Some of the new users on the site. And so now we have Google Gemini Assistant that helps a new user tell them this question has already been asked. Don't even ask it. Here's the answer to it. You know what, why don't you word it this way? And that's actually like where the question was trained off of our data. So that is one thing that we've done in the published slide. The other thing that we did is we launched something called AI assist. So if you go to stackoflow.com today, you'll see on the left NAV when you click on AI assist, it gives you like a very clean kind of, you know, which allows you to ask any question in any sort of natural language. You don't have to be super precise like you have been historically with Co flow and you can ask it and basically it's going to pull from, it's going to be a Gen. AI answer. So it's going to leverage LLMS. It uses Open AI in the background and then produces an answer from not only the corpus, the large corpus of information on Stack Overflow, which is again about 15,000,000 questions and answers, which roughly is like 60 billion tokens of information and also the world's information that's being leveraged through Open AI functionality. But not only does it produces the answer, it gives you the answer in a summarized fashion, just like you know your typical Gen. AI output now, but it gives you the deeper links back to the Stack Overflow con. You can click on them, go deeper, learn about those things much more, and none of that helps. You can still post a question straight from there into our QA area. Or most recently we also launched an ability to post it into our chat rooms where there are experts like live chat. So people are, you know, they're icon chat rooms and JavaScript chat rooms and all these and all the experts are in there and you're able to then engage there and that's so you're able to do that as well. So that is those are a couple examples what we're doing on the public platform. And maybe one other point I mentioned is in order to drive real connection and learning etcetera, beyond chat etcetera, we have also launched clinical code challenges where people are able to learn various things and we post score challenges every week and people are able to continuously learn new things by it's very basically like a hackathon or you can you know, it's the ability for you to learn a new technology and that's also sort of get people on their journey of progressing in their, in their learning and whatever technology. That brings more and more people into the site as well. And of course, they get to your earlier point, what's the most extrinsic motivation? They get recognized for their for their expertise that approves to their to their profile. And then that profile can be used to go get a job and so on. So you know you need more what to learn, you can learn from your fellow community members, you can get smarter and then hopefully you're well prepared for the future, which is going to be like this AI heavy future. And that all fits into the overall mission of being like the vital source of like accurate information, that is. Correct, that's exactly. All fits together, yes, the AI has been like really rough waters. Now it's like a hard to figure out like what's actually going to happen 10 years later. So so like, did you have multiple plans that you were choosing between? Like how do you decide like what the right answer is? So much is unclear. Yeah, it's an excellent question. There's definitely an element of like staying true to the problems that we need to solve for our users and for our customers. So we're constantly asking that question over and over again. OK, New development has happened this week. GPT 5 just launched earlier today. With that, what are the new user problems? Are they any different from the ones we were solving before yesterday or are they any different? So we have to ask those questions constantly. Needs to be rooted in a pretty big problem that people care about and that needs to be solved and that nobody else has solved. Or we can solve it better, fast and cheaper. And that's the core of the question always in either the community product side or in the enterprise product side. And there are different problems for each, right? Inside companies, it's all about everyone wants to use AI, but nobody's really using it really to it's full potential in production and running it in production, etcetera. Because there's concerns about data and security and accuracy and reliability and all these things which are all like make make a lot of sense. So we can help with our platform because we're like this knowledge, intelligence infrastructure and layer that's called this human creation of experts and so on. On the public platform, it's, you know, OK, so you can now generate code. So what do you do there? You can OK, so you need to connect. Hence, let's cultivate community between experts, fellow experts, you know, and experts that's people need to conditionally learn. What about people who never use AI tools? What are they going to do now? You know, they better learn because if you don't use AI tools, you'll get left behind. That's also true. So how do we help them learn as part of that? That's a problem to be solved. And there's also, of course, how do we unlock their growth, career growth and how do we make sure we can prepare them for the next level? How do we get them into and we have this great partnership with Indeed, the job board company where we've got in the world's biggest AI jobs. The, you know, all the developer jobs are all posted there on stackoflowjobs.com which is off of our Stackoflow site. And that is over time we will see that unlock based on people's ability to learn new things, right. So people will be able to apply for certain types of jobs over time. So all those things are problems that we are rooted in the problems and we're constantly asking that question. Our product teams are. And ultimately we've had to experiment quite a bit to sort of keep to understand and kind of test our hypothesis, Yeah, on both sides of the house and ultimately keep iterating based on those early experiments. I've mentioned a lot of things that we're doing at the moment, but before this point. So all those are, I would say, things that we have heard from customers that yeah, solutions to their problems. Yeah. How do you set these? Like, like, how do you set these things up? Like do you have like Advisory Council customers? Do you randomly pull like users on the website and e-mail them and give them like a $10.00 gift card? If they answer the question? Do you have focus groups? Do you hire those big marketing firms to get more data? Like, well, how do you keep a post of what's happening? Yeah.

[32:09] How to keep a pulse on new AI trends, and experimenting new featuresI would say everything except all of the items that you mentioned except for the last one. We don't hire a lot of marketing firms because a lot of data, but because there's obviously plenty of data available. We have our own data, data source. We replace that last point with our own developer survey, which we conduct every year and it goes out to, you know, let's call it 5060 thousand people and people respond rather and there's a tremendous amount of information. You should go check it out. By the way, it's on our website. That is, I've got some tremendous information on people's needs in the AI space, their their careers, you know, what programming languages, what they need help with, what are the issues in that the companies like all sorts of topics. And that gives us a tremendous, it's open and available for everybody to use in the ecosystem. We do that every year in the spirit of community. And but we absolutely learn a tremendous amount in that that survey as like our, our kind of pretty big kind of our data set set of data points. And of course, we conduct our own research groups, product research. We have a team that does that. We have product managers speaking to customers all the time. We have pilot customers that are, you know, building stuff with us. What I was describing previously, the knowledge ingestion as well as being able to plug into your AI assistance. We're building something called a model context protocol server or an MCP server on the top of our products. So that's basically stack and tunnel with knowledge ingestion and MCP on the top of all that is being built with our enterprise customers who opted in to wanted to do that with us. They pulled us in that direction by the way, they came and said, hey, we're building this already internally, we'd love for you to do this. So 100% all we have to build with our customers and with our community always such correct. You know, like experimental wise, like do you have like like like skunk works where out there there's like 5 engineers who are just working on like different projects. You have hackathons where you have like 1000 new projects. Like how do you do experimentation? Yeah, excellent question. In different times, we've had different things. I think that, yes, we've absolutely had the skunk work teams. Back in 2023 specifically, we had to carve out a team to go launch those AI things that I mentioned previously, which is called overflow AI, et cetera, you know, Enterprise Products. And we had to carve out a team, we carve out like 10% of the team to go product engineering team to go and focus on just that, right? And we're just going to double down on it. And and then of course, it has become now more mainstream work. At that point, it felt like, OK, we got to go carve that out. Now it's obviously the work, you know, the foundation, but we also encourage very quick, we have certain sort of forcing functions in the company where we have, for example, every couple weeks we have weekly business review. So we always talk about what's been shipped over the past couple weeks. And depending on what we're hearing from users and customers about what's been shipped, we may decide to go either in a different direction or keep going down, but in a slightly modified way, or we may decide not to proceed with something based on some input we've done. We've seen all those over the over the years, right. So, yeah, it's, it's constantly because, you know, to your point, there's so much changing and so it'd be foolhardy and I think naive to think that your first attempt is going to be the correct attempt always. I mean, it's not, you may get Supreme really happy and that's great, but I don't think you want to count on that. So you want to have, I'm always a fan of making sure that we have a spirit of experimentation and we have we try a whole bunch of stuff until we see what ultimately works for customers within that problem. Gotcha. And then back in 2019 when everybody decides to go to like Enterprise, that seems like very brilliant looking back to it, Like, how did that decision come into play? In many ways, I think it was the the signal that they had received from the customer base even prior to me joining the company. So it was some of these large tech companies like a Microsoft and I can rattle off a whole bunch, but I'm not doing I'm really allowed to share the names back then all started saying, Hey, we love Stackoflow. We just love the private version. So literally the company made a copy of Stackoflow and gave it to them and they use it inside the company, right. But it wasn't like that was the earliest version of this aspire. But over time, it became obvious that this could be like a real product that companies would side. And so that's when it was, that's kind of when they hired me because my background was about scaling enterprise businesses and recurring revenue businesses at companies like Rackspace, which is here in Texas. And that expertise was useful in that context, right? To effectively effectively transform the company from AB to C or even B to D business to developer community company, AB to B software company.

[36:35] Financing for the companyYeah. So on top of all the technology, there's also like a financial like how did you decide like when to go out for more financing? How do you decide the structure of the company over the years like? Yeah, it's good, good question. So I joined in October of 2019 and in March of 2020, about six months after I joined was COVID. And that was quite because, you know, ultimately the, the, the job market, if you remember every stop hiring, right? So the jobs business went down by 50% six months joint. It's like welcome to the company. Advertising stopped because it was like, you know, went down 50% because it was, it was advertising. Why, why spend money on marketing when like the world has been shut down because of coal? So there was like this kind of like this shock moment in the company. And so we use that as an opportunity to really sort of lean into the future. So we said we're going to go to the enterprise anyway. We might as well do it now. So we actually shut down the talent job listings business, which was, by the way, 70% of the company's revenues at that point. And so that was fairly unnerving to do. But we had the support of the Board to go do it and we raised a new round of financing in 2020 with some amazing investors came in, in 2020 to help us invest into the enterprise business. And that was like an $85 million round, Series G round we raised. And that was really instrumental in getting us the ability of the kind of the firepower to go and bring on the right kinds of people onto the company, the sales reps, the customer success folks, the demand generation leaders and the people and really sort of build this motion into a kind of well oiled machine. That then sort of took some time to get going. And that started really then scaling the business really quickly. And that's, that's ultimately what led us to our to us being acquired by our current owners or process or also known as Naspers in Europe. And you know, so we've been a process portfolio company and their growth equity type companies where effectively now in this AI journey with them with this like how do we go from being a SAS company to an AI company, right, a company. So that's basically what we're up to doing at the moment. So with the VC, you still sort of had more control, but as part of a portfolio company, you have less control. How do you make sure to find the right partner for the Cycle virtual business? Yeah, I think it's a good question. I think it's like ultimately you have to be aligned in like the the vision of where things are going and you know why you bring brought on, right. Investors are very important stakeholder in this company. You've got obviously like every company has got customers and investors and employees and this company, we have a huge community that's the 4th stakeholder. So it's an even more complex, but I think investors, it's very important that you're on the same page with them because ultimately if you're aligned, there's nothing. It's pretty amazing because like ultimately what you're trying to do is accomplish an outcome of some kind either. Everything is rooted in building a great business. With that comes the ability of course to earn a financial return. And for the first set of investors who that brought us, that was like the when I said we got acquired, it was a great outcome for them because we were acquired a great, a great number back in 2021. People were very, very pleased with the outcome. So now, now we're on the second journey like it's another sort of post first acquisition sort of journey. That's where we are. Yeah, on on your personal side,

[39:57] How to become a great CEOhow did you become a good CEO over a time after studying Masters of engineering management at Cornell? Yeah, yeah. Well, I think, I think Cornell actually, specifically the masters of engineering management, I did my, my undergrad was in computer engineering. And you know, I knew early on with a couple of my internships that were in engineering that, you know, my my interest and just my inclination was a lot more into towards management topics or people topics. And I sort of enjoyed thinking about how the companies scale and how do they become impactful and you know, what is the machine and it could move macro level that you need to put in place to sort of make them successful. Those were more interesting questions for me than, you know, writing code even, which is ironic. I do, but but I, but the fact is that that was, I could do it. I could do it well. And that's what I started to do, but I was even more interested in like, how do you create like a system? How do you create kind of an organization? And that took me on the journey of going to Cornell right after undergrad and that it was an amazing experience. I mean, it was, I still remember like, you know, the image program I think is so unique because it allows you to sort of immerse yourself in the stuff you're not exposed to. For example, it's half engineering, it's half kind of the school all wrapped in one. And you get to do like a Co-op with one of the big companies and how we did ours with the General Motors, right? All about operations efficiency and the manufacturing facility, just like an amazing experience here. I got to lead A-Team at Cornell as part of that project and just opened up things that I'd never really thought about about myself. And so I think it was like a really sort of formative period of time. And I just like to absolutely love my time at Cornell. It's such an amazing campus and welcoming, welcoming environment, amazing professors who I still keep in touch with them. I served on the engineering boards there for multiple years until recently. So that I think there's been just like, I think lifelong journey of moving in that direction. I think for me it's always been about becoming a lot more well-rounded. And maybe it is based on my upbringing with my parents who always encourage that to sort of be well-rounded. My mom always said, you know, let's put you in athletics, let's put you in make sure you learn a musical instrument or two. Like, and that I think exposed me to like wanting variety and wanting to really being energized by having that well-rounded view as being super deep specialist on something, right. And I just like built that way. Like I enjoy like the broad purview. I think that comes to my dad, who also loves the sort of the broader purview. And that's just settings like this natural, like interest. And that's that's what took me to do that. Go to Cornell, go to consulting, go to Business School after that to kind of found out things like strategy and finance. Go to banking and work in again, technology telecom company, focus in the bank and then work with a tech company running up in an operating role and trackspace here in Texas leading several high growth businesses and really being a general manager and sales teams and product teams and, you know, engineering and operations teams and doing it sort of end to end and understanding what does it take and sort of to run a small company. And then, of course, when I got the opportunity to do exactly that here at Stack, I was thrilled to take on the challenge. And CEO like the broadest thing you can do. Right. I think the big difference between doing it inside a company as a general manager and doing it as Ceoi think is of course, it's the fact that, you know, buck stops that use, you know, as so you've got nobody to sort of turn to other than yourself. So you got to figure out what to do and you know what the base, where the base should be, what decisions you need to make. It's all on here. So versus the large company, you have the little bit of the luxury of looking up or looking at, you know, for other sort of guideposts to say, oh, that's kind of what we're trying to do. Let me like move in that direction. So in this case, it's, you have to figure it out based on based on your own like, you know, intuition and just what you're trying to accomplish. So for the closing question, I

[43:47] Closing questionalways ask the guest, what's the kindest thing that anyone's ever done for you? Yeah, easily. It has to be some of my earliest managers who put a tremendous amount of trust in me in very high stakes situation situations. And I look back on that as being like, this is tremendous, like learning experience. Like I would have not been able to have learned the things that I did that early on in my career if it were not for some of these managers who truly extended trust. And, you know, whether or not they did that with nervousness or not, I'll never know. But they just sort of, they just did it because they put me in front of very senior folks within some of our clients and I specifically in my consulting days. And it was like amazing to be leading this kind of extremely experienced folks around the room on a topic that I may have had like raw, like horsepower on based on, you know, just like, you know, understanding it. But and it was about like, how do you like lead this group to get to a certain outcome? And now, you know, they're going to listen to you. As you know, it was like a young 22 year old when that happened, right? So like for me to be able to be exposed to that sort of responsibility super early, like I'm ever, you know, very grateful to those folks who gave me that chance. My earliest managers, I think they'll I'll always remember them for that. Thanks for sharing. Yeah. Thanks, Tony.