Prashanth Chandrasekar Transcript
Clint Betts
Prashanth, thank you so much for coming on the show. Means a lot to have you here. You're a big-time CEO. CEO of Stack Overflow, which is an incredible company. It might be useful for those who don't know, and I would be surprised if people don't know Stack Overflow, honestly. Who doesn't know about what your company does and how you became the CEO? If you could give us, maybe we could start there.
Prashanth Chandrasekar
Yeah, sure. Stack Overflow is the world's largest software developer community. We serve about a hundred million people from all around the world. And we've been around since 2008, so it's about 16 years going right now. So, we are very privileged to serve folks in about 185 countries. The reason why we're so popular is because we have the world's largest knowledge base of technology information. So, we have close to 16 million questions and answers for every possible programming language and scripting language, etc. And that's the foundation of the heart of the company.
Beyond that, of course, we have several products that we sell to companies, including an enterprise version of Stack Overflow that's used as a very accurate knowledge store of information and knowledge within companies. And we now have close to 20,000 customers in that space. And then we have an advertising business because we have so much traffic from our website. Most recently, we have our AI products, which are called Overflow AI, which are generative AI capabilities that sit on both the public and private versions of Stack Overflow as well as something called Overflow API, which is all about providing our data set and knowledge base as a service to companies and partners so they can use accurate information and things like model training and other AI use cases.
Clint Betts
How did the company look in 2008? Was it forums? Is that how it got started? Almost like a blog and forums?
Prashanth Chandrasekar
Yeah, great question. I would definitely not describe us as a forum, even though it's a natural place to go. Our founders came together with a very small group of people, saying, "How do we democratize access to knowledge on all the things that people are frustrated about when writing code?" Including myself, I never had Stack Overflow when I was writing code back in college in the late '90s. And for those who actually wrote code back in the day, it was extremely frustrating. You're sort of on your own having to figure out, "Oh, I missed a close bracket somewhere." And you're pulling out your hair until odd hours of the night in the morning.
So I think that was the nexus of coming together in a small but growing, very rapidly growing community, which is like, how do we all work together to put all the common issues that people are encountering to help each other out? Truly based on the volunteerism of the community members, it grew from a small number of Q&As to 60 million questions and answers. The reason why it's not a forum is that it's not an opinion or a discussion board of any kind. It's very much an accurate knowledge base of the canonical answer to a question. There may be multiple answers to a question, but it's always peer-curated; it's voted by the community. And only the accepted answers are ultimately displayed for people to use in a repeatable sense. So it's not a discussion; it's not like a podcast conversation like this one that just shows up online. It's very much the equivalent of an encyclopedia of information that's very accurate and peer-curated by the world's experts.
Clint Betts
At what point did you get involved?
Prashanth Chandrasekar
I joined the company in 2019, so almost 11 years after the founding of the company by two amazing founders, Joel and Jeff. They brought me on because the company's phase around 2018 and 2019 involved a couple of businesses. The advertising business had something called a talent business, in which we run job listings, but then they realized that there was an opportunity with what I described earlier, which is the enterprise version of Stack is this problem that we're solving to make sure that people have accurate knowledge inside companies. Those are primarily the issues of finding information that's accurate within companies. It's always a big productivity drain.
So they thought that there was a great opportunity for us to lean into that. And they wanted to bring on somebody that had the ability to scale into a recurring revenue business, et cetera. The work I'd done in my prior role at another technology company here in Texas was all about scaling and growing a very rapidly growing division of a large company. That is why they found me, and I was thankful to have the opportunity to make a difference here.
Clint Betts
I mean, this question is pretty broad and probably takes hours and hours and hours to answer, but how do you think software development has changed since Stack Overflow was founded?
Prashanth Chandrasekar
Yeah, I think it's a great question. I mean, there have always been moments in time where it was the pre-and the post. It was never the same after those moments, whether that is the internet era, the mobile era, or even the cloud. And then Stack Overflow, if you think about the world before Stack Overflow, where people like me who were writing code based on my own intuition, plus my knowledge, plus maybe asking people that I knew around me to help me if I got stuck, or maybe asking a professor if you were in university. Those were the pre-Stack Overflow days, obviously documentation and manuals, and all the things that we typically use.
Post the founding of the company in 2008 and Stack Overflow with all this knowledge that started to get documented, it was as simple as people onto, let's say, your favorite search engine, let's say Google, and typing in how do I spin up XYZ server? Typically, the answer was always a Stack Overflow answer that came as the first answer because it was always accurate and the fastest response. And people use that as a very repeatable and accurate metric. You could trust it because it came from the world's experts who had basically upvoted the right answer and downvoted the wrong ones, so only the right ones you knew you could rely on. That became the way in which people started writing code. It was less struggling with documentation and textbooks. And now people have been doing that now for almost over... Since 2008, so for 16 years.
And now we are pivoting to the AI world where we're basically saying, "How do we deliver the same high-quality knowledge in the form of a knowledge store to companies and to developers all around the world, wherever they are?" If they are going to be spending time in a GenAI tool, awesome. Let's surface it there. Let's surface it inside an IDE where they're spending a lot of time writing code, or let's surface Stack Overflow content in your Slack instance because you are spending a lot of time chatting with your teammates, or maybe in GitHub Copilot because that GitHub Copilot does a great job of generating code. Why don't we provide all the context that supports that code that's being generated so you understand what you're actually generating, and you can learn from it, and you can debug it, and so on?
So we just want to be wherever the developer and technologist are. And so our model has changed from relying on people to come through, let's say, a search engine response to our website that certainly a lot of people do that, continue to do that, millions of people, but we also want to be in all these GenAI tools, which is really why we've announced all our partnerships with OpenAI and Google and soon GitHub and others that we're excited about.
Clint Betts
What is your sense of the future of AI? Another huge question.
Prashanth Chandrasekar
Yeah, a few of us have been through, as you have, through multiple of these technology transformations. There's always going to be this adoption curve, and we're very much in the early parts of that adoption curve. If I go take it back to some data, we always conduct a Stack Overflow developer survey every year, and we reach out to call it 60,000 to 100,000 people a year who respond. Last year, the inclination for people to use AI tools was, about 70% of our community was like, "Hey, I'm really interested in using it," or, "I plan on using it," or, "I'm already using it." That was 70%. That number went to 77% this year in our survey that we just concluded.
The trust element of how much people trusted what was coming out of the AI tools was around 40% last year and was exactly the same percentage this year, 40% a year later. So that hasn't changed either, right? Which is like there's high inclination to plan on using it, but the inclination to trust the tools actually was not very high. And primarily because people are concerned about accuracy. They don't know where this information came from. It's a black box. It does not cite any of the original sources. Who knows? All those things are obviously concerning when you think about rolling applications into your bank or any other organization that you're part of.
But the bigger thing that I thought was quite interesting is that the enthusiasm with AI actually dropped a few percent. Last year it was about 70%. There was another metric that we tracked and that dropped by about 7 percentage points. It was pretty high, and it is a meaningful drop. And primarily because I think people have begun to realize the realities of rolling out AI into, let's say a company. So all the points I just mentioned, trust and accuracy, and "Who's going to create the knowledge if humans don't create it?" And "Who's contributing to the accuracy of these models?" And so effectively, some of the shine has come off of the topic as people go through the hype cycle, but there is certainly still, I think, interest in forward movement on the topic.
So where we are in summary is that I think people have realized that you want to have humans in the loop for the foreseeable future to make sure that you are not only attributing and you're able to trust the information that's in there but you're also creating mechanisms and incentives for people to contribute new knowledge. So, these models can train off of new information. These models absolutely require new information and real-time information to keep it up to date.
Then, ultimately, the AIs don't really have all the answers to all the questions. Either they just don't know the answer, or they hit a complexity cliff very quickly. They're great for early drafts and those sorts of things, but certainly, it's a cliff. At that point, we don't want to lose the user as we do with these AI tools. So why don't you leverage a community like Stack Overflow to complete that workflow? That's how we are engaging with companies. What we're calling knowledge as a service is to continue to keep that loop going so the user always gets what they want and gets back into that flow. That's where we operate as part of the AI ecosystem.
Clint Betts
How often are you using AI tools? Just personally?
Prashanth Chandrasekar
I don't code anymore as much or very much at all, but I do obviously use... I test a lot of things on my own, including our own products. We also have, for example, Stack Overflow for teams inside our company. So, I use that quite extensively. It's called Great Semantic Search functionality, which we've now launched with our customers in our Overflow AI module, which customers now leverage. So, to find information inside our company to onboard new people inside our company, I use that all the time to respond to questions and automatically find things within our own company.
And then that's within Stack Overflow, but outside, just on my own. I would say, in my personal time, I have found open-source AI tools to be quite interesting. So, for example, Meta's Llama 3.1 is an example, a good example of something. And the moment that it was launched, I thought the way they distributed it was very clever. And WhatsApp because I use WhatsApp with my family all around the world as an example. And for me to be able to just engage in that form factor, similar to my earlier point, we want to be wherever the developer is. In this case, I'm an average user on WhatsApp, and I'm able to leverage and engage with an open-source AI model, which I thought was very clever.
I use that quite a bit. I mean just mostly to various things. It could be as simple as me educating my parents about XYZ topic and giving them a summarized easy-to-understand way. It could be created in the form of an ancient poet who has a very specific style of speaking. So those sorts of use cases. Nothing mission critical from a personal standpoint because I think we still have a long way to go to do things like complete the task of all the things that people have talked about, "Go and book my vacation and XYZ." I think that's all a little bit away. But yes, internally, in the enterprise, I use it for searching, finding information, and making sure that we're automatically capturing knowledge inside a company. I use that heavily. And then, on my personal side, it's the search equivalent of open-source tools.
Clint Betts
Zuckerberg has really become a hero for developers, such as open-sourcing AI versus OpenAI, which, given the name, you would think was open. It's actually a closed source. Do you have any opinion on open versus closed? Obviously, there's always going to be both, but do you have an opinion on which one would be better for developers to work on?
Prashanth Chandrasekar
Yeah, I mention one thing about Zuck. I don't know him, but just watching from the outside, I would say it's a masterclass in rebranding both yourself as an individual and the company. I think what I've observed over the past several years is absolutely phenomenal. The fact that they are now the champions of open source, especially in the AI world, which is the biggest technology movement of our time, is just absolutely phenomenal to see.
So, in terms of open source versus closed source, I mean, given who we are, we're huge believers in open source. Of course, we want to make sure that people have access to information, including... By the way, even in our overflow API program, we're actually licensing our data to these big AI providers like OpenAI, Google Gemini, and so on. The data that we have for our average community user is always open, accessible, and free, and they can use it for all those things. We obviously engage with these companies in commercial contexts because of the integrations that I mentioned previously. So I'm generally, and our company generally, and our community generally is very much open source. So that's the first point.
The second point, I think, is that it's becoming very obvious that the LLM wars that are underway, closed source or open source, are all headed in one direction, meaning they all want access to all the latest data. So, data is absolutely a premium. High-quality data, especially sources like Stack Overflow, which is one of the few places where these LLM models are trained, are absolutely important. So that is one. Over time, you can imagine that it's an expensive proposition because of the cost of chips, the cost of infrastructure, the cost of energy, and all the things that we've heard about. And that invariably, currently, that cost is being passed on to users and customers through your a hundred dollars per month or whatever the charge is for any of this combination of AI tools. At some point, that cost will go down. History just suggests that it's going to go down. And I think there will be a commoditization of things like search functionality. GenAI allows everybody to have search functionality, which is interesting.
So I think it's going to be hard not to have cost pressure for these things while they're expensive for that cost to go down. Hence, I am quite bullish on the open source ecosystem for AI because, especially with Llama and Meta's move, I think it's, again, a masterclass chess move to be able to do that. Because they're probably one of the few companies who can actually do what they're doing, massive amounts of compute, massive amounts of GPUs, huge amounts of data, and for that to just be open source, it sort of takes the rug out from all the other folks who are closed source.
So yeah, I think this is going to be a fascinating thing to see, but I don't think there's a victory yet, just to be clear with what I'm saying. I do believe it's going to be fascinating to see how each of these iterate. There are obviously things like safety and privacy, and this is a better model. Is a closed source model better for safety and privacy, even national security, to a degree, versus open source? So there's all sorts of pros and cons. On the margin, given who we are, we are big believers in open source.
Clint Betts
What does a typical day look like for you as CEO of Stack Overflow?
Prashanth Chandrasekar
It's one of the joys of being in roles like these, which is no day looks the same. I mean, it's absolutely dynamic. It depends on what the most important and the most urgent thing is, right? It's kind of like that two-by-two matrix. You want to focus on the most important, most pressing, and urgent things. So there's some combination of thinking strategically about the future. That's a portion of my day. It's some combination of making sure that I'm talking to customers definitely on a daily basis. On average, I have at least a couple of customer calls, if not more, a day. And certainly internal meetings in the context of reviewing and driving progress. I think one of the things we have to keep moving with pace and with a sense of urgency is while the ecosystem is innovating, are we innovating at an equivalent pace? So I think it's all about whether we are driving towards our goals, which are all the usual things that CEOs do. That's the third piece.
And the last piece, of course, is the humanistic element of staying accessible as a leader, making sure that people understand across the company, "Look, here's where we're headed. Let's just have an open conversation questions," just to make sure people are coming along for the ride. Because there's a lot changing, and it's ultimately up to leaders to clarify, reiterate, and repeat. One of my other titles other than CEO is CRO, not Chief Revenue Officer, but Chief Repeat Officer. So, I constantly mention the same thing over and over again. And that just... I have to do it just to keep people aligned. So that's probably, on average, I would say, the chunk of my day. The four pieces.
Clint Betts
What are you reading? What reading recommendations would you have for us?
Prashanth Chandrasekar
I like to think about a couple of foundational books, and I'm not like a guy like Bill Gates, who consumes hundreds of books a year and things like that. I'm not that person, but I do read a lot of in-the-moment content that I'm very curious about. I'm naturally a curious person. So, I'm naturally reading, especially as it relates to AI. These days, there's so much content. Even for most, I would say, well-read people and people who keep up with that sort of thing on a regular basis in the technical community, it's quite daunting to keep up with the developments on a weekly and daily basis. So I find my spots on... I have a couple of subscriptions to things that keep me updated on the latest in the AI space. And beyond that, the foundational books are all the things that I'm happy to share. But they're examples like team books, for example. I'm a big fan of Five Dysfunctions of a Team. It's a great Pat Lencioni book.
Clint Betts
Yeah, Patrick Lencioni. Great guy.
Prashanth Chandrasekar
Indeed. And the other one would be Carol Dweck, The Growth Mindset. It's also an amazing book that I read quite early in my career, and I've found it to be quite... I mention that all the time to people I engage with. So those are probably a couple of foundational ones, along with all the usual suspects of business books that you probably know about, Good to Great, Speed of Trust, and all the other ones that I can rattle off.
Clint Betts
As a leader, what do you think are the three most important traits for one to possess if they're going to be a leader?
Prashanth Chandrasekar
I mean, I think the definition of leadership is very personal to people, especially individual leaders. So there's no one answer to that question. I'll tell you from my perspective, I think it's very important for leaders to bring a level of energy and vision to wherever you're trying to take the company. You need to be very clear and energetic about where you're headed because if you, one, don't know where you're going, that's a problem because people just don't know. They can just do a lot of stuff and just go around in circles. So, I think one is setting that vision.
Two, really having forward momentum and forward energy, and not just saying it and talking about it, "Let's go fast." It also operationalizes that energy within the context of the company. And I think that is a second part: how do you have that forward momentum? And the third one is it's all about really energizing, I think, the populace to actually all move in that same direction. So it's almost like three parts of where you are going, making sure that you can drive with energy and ultimately lead by example. Say, "We have to go with speed." Especially relevant in a company that's, let's say, in the high-growth tech space, et cetera. The third one is how you really motivate people and really energize them to follow in that direction. I would say those are the three pieces.
Clint Betts
What do you think of this new trend? I mean, it's been going on for a while, but it is relatively new in the history of business that people look at CEOs and leaders to speak out on things outside of their company. You have an opinion on macroeconomics, political races, what's going on in foreign countries, and wars. How do you think about that, and how do you manage it?
Prashanth Chandrasekar
Yeah, it's a good question. I mean, this has obviously been relevant over the past several years as so many things have happened since the pandemic or even pre-pandemic, right? It's been happening for a while. I think that CEOs have a responsibility to acknowledge what's going on around the world. And I don't think it'd be really authentic as a leader if you did not acknowledge it. And that's important because these events, while they may not be directly relevant to your company, affect people's lives. Ultimately, when people are engaged in your company, they spend, let's call it whatever, 10 hours or 8 to 10 hours a day at work a day. But that's out of a 24-hour day.
So, in the remaining time, they have to think about their family and friends, what's going on around the world, and how that impacts them. It's their taxes. There are wars. There are relatives abroad that were impacted. There are elections that they're curious about, whether their state's or blue. There are so many things that are part of their lives that impact how they show up to work. So I think that it would be really obtuse for leaders to say, "Look, I actually don't care about any of those things, and you have to do your job here." So there's a balance, which is the main point I would make. I think acknowledging it gives people space to engage in a constructive discussion. And not necessarily in a company-wide setting because we have a mission to accomplish, which is important to focus on. That's the primary people, the primary reason people are at work. But if people want to have the ability to share that, there are plenty of avenues that I think companies should provide.
So, as an example, we have a number of ERG groups or employee resource groups that talk about various subjects. And they're all led by employees. These are safe spaces for people to go and talk about various things and really have the ability to engage in constructive discussion and have the ability to share their thoughts on it. Invariably, some of those things will connect back to things like how we think about what we write, what we need to be sensitive to, what's happening externally, and how we design our products. So, I think that context is important. Then you have a sense of, "Okay, what the world is going through." Everything is interconnected, as we know. So, I think it would be silly to just ignore what's going on. But I also want to be clear: that doesn't mean the company can just spend 80% of its time thinking about all this as a primary focus. The primary focus is to ultimately accomplish the mission. So I think that is the focus. And so there is a balance, but hopefully, I've described to you how I think about it.
Clint Betts
Yeah. How have you landed or how have you managed on this whole debate around work from home, hybrid, everybody in person?
Prashanth Chandrasekar
Yeah, we've had a naturally hybrid company for a long time. So I think about 60% of our company was... For example, our marketing organization was remote, 60% pre-pandemic. 80% of our product engineering was remote pre-pandemic. And overall, as a company, I think we're about 40%. That is only because most of our sales folks were actually in the couple of offices that we had in New York and London. But post-pandemic, we moved away from those offices for all the obvious reasons, and we just became a highly remote company. And we operate as a 100% remote company, probably one of the few. And some of the other companies have... The pendulum has swung back for them to get people back in the office and so on.
We naturally do that as part of, once a week, we have a lot of shared space. I sit in a shared space. Even this office here, I'm sitting in an office. I'm not sitting at home. Only because I find it to be more productive. And so many of our teams come together on a regular basis to do the same. And we work in the regions of the world. Once a year, we come together as a company to do an annual company kickoff. Most years, we've done that in person, and some we've done it virtually. Just sort of depends on what's going on. Hopefully, there will be no more pandemics. Those are scenarios where we've handled that.
Clint Betts
Finally, we end every interview with the same question, and that is at CEO.com we believe the chances one gives is just as important as the chances one takes. When you hear that, who gave you a chance to get you to where you are today?
Prashanth Chandrasekar
Yeah, easily; I would say if I take it all the way back, I am an immigrant story. So, for me, it was; I grew up in India until I was about 17. I had this opportunity to study abroad based on a deal that I made with my parents, saying, "If I got a scholarship, I would do it." They were very uncomfortable about it because I was only 17, and they had never even known anybody leaving home that early to go thousands of miles away, and so on. And so when I got my college scholarship in Maine, I was absolutely grateful. And so they gave me a chance. None of this would've been possible if not for the school that gave me an opportunity to come over from India to the US and make my way over a couple of decades after that. And so I'm very grateful to both the university where I now serve on the Board of Visitors as a result of my attempt to give back. Also, obviously, the United States is the greatest country in the world.
Clint Betts
Prashanth, thank you so much for coming on. Seriously, it means a lot. You were at the forefront of a whole new world coming in with this AI stuff and how developers work with it. And it's incredible to talk to you. Thank you so much.
Prashanth Chandrasekar
Thank you, Clint. Thank you so much for your wonderful questions and thoughtful questions. So, I look forward to more conversations.
Edited for readability.