Stack Overflow users don't trust AI. They're using it anyway
Decoder with Nilay Patel · 15 Dec 2025
Source: https://www.podtrac.com/pts/redirect.mp3/pdst.fm/e/pscrb.fm/rss/p/mgln.ai/e/257/traffic.megaphone.fm/VMP2195361773.mp3?updated=1765577524
Transcript: https://podcasts.opennash.com/shows/decoder-with-nilay-patel/stack-overflow-users-dont-trust-ai-theyre-using-it-anyway-22a174/ (via OpenNash Podcasts, machine-generated)

Support for the show comes from Comcast Business. With advances in AI, cybersecurity is getting more powerful, but so are attacks. In the past twelve months, Comcast Business detected 24.5 billion phishing attempts designed to exploit user behavior, steal credentials and compromise devices. AI is forcing a shift in how organizations defend themselves, demanding not only faster responses, but continuous intelligence. To survive in the current climate, leaders must build a security-first culture to equip their team with the tools they need to stay protected.

With Comcast Business, connectivity, intelligence and response work as one orchestrated system. Get better equipped to handle today's cyber threat landscape with the 2026 Comcast Business Cybersecurity Threat Report. Read it today at Comcast Business dot com slash cybersecurity. Cyber confidence starts with Comcast Business.

This holiday, switch to the Simplicity Plan from Verizon. It's one plan, one price, just $30 a line per month for you or the whole family. And on their best five G network, switching to Verizon and shopping their best deals makes the holidays simple. Head to Verizon dot com or your nearest Verizon store today.

Support for the show comes from CrowdStrike. It's not a huge stretch to say that AI is the next major computing platform. Every platform shift changes the way we build software, and it changes security too. CrowdStrike is defining cybersecurity in the AI era with AI detection and response, or AIDR, a solution for seeing, monitoring and securing AI across the enterprise. Learn more about how leading companies are turning to CrowdStrike to secure AI and secure their business at CrowdStrike dot com slash decoder.

Hello and welcome to Decoder. I'm Nilay Patel, editor in chief of The Verge, and Decoder is my show about big ideas and other problems. Today I'm talking with Prashanth Chandrasekar, who is the CEO of Stack Overflow. I last had Prashanth on the show in 2022, one month before ChatGPT launched. And while the generative AI boom has had tons of impact on all sorts of companies, it immediately upended everything about Stack Overflow in an existential way.

Stack Overflow, if you're not familiar with it, is the question and answer forum for developers writing code. Before the AI explosion, it was a thriving, major community where developers asked for and received help with complicated problems. But if there's one thing AI is good at, it's helping developers write code, and actually not just write code, but develop entire working apps. On top of that, Stack Overflow's forums themselves were flooded with AI generated answers, bringing down the quality of the community as a whole.

You'll hear Prashanth explain that it was more or less immediately clear how big a deal ChatGPT was going to be, and his response was pure Decoder bait. He called the company emergency, reallocated about 10% of the staff to figure out solutions to the ChatGPT problem, and made some pretty huge decisions about structure and organizations to navigate that change. Three years later, Prashanth says Stack Overflow is now very comfortable, primarily as an enterprise SaaS business, which provides AI-based solutions that are tailored to different companies' internal systems.

Stack Overflow also operates as a big data licensing business, selling data from its community back to all those AI companies, large and small. That's a pretty big pivot from being seen as a place where everyone can go to just get help with their code. So I had to ask him, does Stack Overflow even attract new users anymore in 2025, when ChatGPT or Claude Code can just do it all for you? Prashanth said, Yes, of course, and you'll hear him explain that while AI can handle simple problems, for thorny, complex problems, you really want to talk to a real person, which is where Stack Overflow still brings people together.

You'll hear us come back to a single stat in particular. More than 80% of Stack Overflow users want to use AI or already using AI for code related topics, but only 29% of them actually trust AI to do useful work. That's a huge split, and it's one that I see all over in AI right now. AI is everywhere in everything, and yet huge numbers of people say they hate it. We hear this feedback in the Decoder inbox and the comments on The Verge and on our videos on YouTube.

Everyone says they hate AI, but the numbers don't lie about how many millions of people are using it and apparently driving some benefit. It's a big contradiction, and it's hard to unpack. But Prashant was willing to get into it with me, and I think you'll find his answers and his insight very interesting. Okay, Prashant Chandrasekaran, CEO of Stack Overflow. Here we go. Prashant Chandrasekaran, you know, CEO of Stack Overflow, welcome to Decoder.

You're wonderful to see you again. It's been, uh, as you said, a hot minute. Three, three years, I think, is the last time when we spoke. So, uh, great to see you again.

I should have said welcome back to Decoder. You were last on the show in October 2022, one month later, ChatGPT launched.

That was an interestingly timed, uh, interview right before the world changed.

Right before the world changed. Software development, uh, certainly the thing that has maybe changed the most, uh, since the the AI models have hit. There's a lot of new products in your universe to talk about, and there's what Stack Overflow itself is doing in the world of AI. So I want to talk about all of that. But first, just take me back to that moment. We had spent an entire conversation in 2022 talking about the community and moderation, uh, how you were gonna build a funnel of people like learning to code, learning to use Stack Overflow.

That was a big part of our conversation. The pipeline of of engineers, both learning to write software and then be a part of the software development community, that was very much on your mind. And then all of software development changed because of the AI tool. So just describe that moment for me, because I I think it contextualizes everything that happened afterwards.

It was definitely a very very um, you know, surprising moment, I don't think an unexpected moment in in many ways, because here comes this technology that obviously, uh, some people knew about, but not not in a way that obviously captured everybody's imagination using this, you know, beautiful interface. Um, and uh, you know, I was, uh, in, we were in the middle of wrapping up our calendar year, uh, and at that point it was, here we go, we've got, you know, we would think about our priorities for the next year.

And it pretty, uh, pretty much came to, uh, it would become very clear, you know, what we needed to focus on, because this is obviously going to be this very, very, um, uh, huge, uh, change to how people consume technology. And this is, you know, welcome to technology, you know, it's constantly changing and things, especially this wave, uh, I think it's completely unprecedented. I don't think there was any sort of, um, analogy or any other sort of, uh, you know, prior wave that I could look to, uh, including the cloud and maybe the internet, but, you know, I I don't think, uh, you know, we're still sort of fully sort of consuming what that is the moment.

But I would say, yeah, so we had we went into what is the equivalent of a code red situation inside the company. Uh, it was an existential moment, especially for our public platform, uh, because the primary, uh, you know, the jobs to be done, if you will, is is all around making sure people got answers to their questions. And here you go, you have this really, really slick interface, that natural language interface that allows you to do that, you know, in a moment's notice.

So we had to sort of organize our thoughts. And uh, what I ended up doing was carving out 10% of the company's resources to very specifically focus on a response to this. Um, and we set a very specific date to respond by in a meaningful fashion. So we said, hey, the summer, uh, of uh, 2020, uh, three. I was gonna go speak at the We Are Developers Conference in Berlin. And I effectively told the company, hey, we've got six months to go and produce our response, at least our initial response, because obviously this is gonna keep iterating and so on.

And that's how we mobilized the company. We had this, uh, we acknowledged it was a code red moment, uh, we carved out a team of 10%, you know, so that was about forty people or so. So, you know, we're uh somewhat of a a medium-sized company. And then we we got to work, and that that was the moment.

Take me inside that room. Very few people ever get to send the memo that says it's a code red, right? This is not a thing most people ever get to do. I mean, maybe think about doing it, but maybe no one's gonna read your memo. Everyone has to read your memo. You're the CEO. Take me inside that room where you say, Okay, I have identified an existential threat to our company. People have come to us for answers to software development questions.

Again, the last time we were on the show, you were talking about the idea that there were objective right answers to software development questions, and that the community could provide them and vote on them. Well, now you've got a robot that can do it and can do it as much as you want, as long as you want. And now with tools like Cursor, it can maybe just do it for you, right? With tools like Claude Code can maybe just run off and do it for you.

Okay, so you've got that, and you say, I need to take 10% of the company. I'm curious how big the company is. I know there's been some changes, but 10% of the company is just forty fifty people. How did you identify? Okay, this is the moment I need to pull these people in the room. I'm making this decision, and the right answer is forty or fifty people are gonna set aside their time to deliver me a plan by the time I give my next keynote.

the instinct to do that was, um, you know, it's come from a couple different experiences, you know? Uh, my experience right before that was at Rackspace, uh, in the cloud services space. And, you know, the business I was actually running at Rackspace was all around. How do you respond to Amazon Web Services as a cloud technology threat? I was in the team that was, uh, ultimately rebuilt that business from the ground up, and that it was effectively the 10% of Rackspace's population that went and, you know, went and created that.

So I had had some practice on, you know, what does it mean to see and respond to a disruptive threat that, you know, you're encountering? So I was, so this is my turn now to sort of put that into motion at Stack by appointing somebody like myself at Rackspace. When I did it, I had to go do exactly the same thing. The other data point was if I go all the way back, you know, a couple decades ago, or more than a couple decades ago, uh, when I was in business school, my, you know, my professor was, uh, Clayton Christensen, uh, and, you know, he wrote the book Innovator's Dilemma.

I go back to that because I have always thought about that in the context of technology, because technology, it is a very consistent theme, uh, leave alone other industries, that every so often you will have disruptive threats, and there's a very specific way in which you need to respond to that. You know, the history suggests that you should carve out an autonomous team, uh, that has very different incentives and can pursue things in a very different way relative to the rest of your business.

And remember, our s our company, Stack Overflow, has really two parts. You know, we have our public platform, which had this sort of, you know, this web, uh, kind of big disruptor, which we should talk about more broadly about the internet. But then the other side of our business was the enterprise business, where we're serving large companies with our private version of Stack Overflow inside the company. So thankfully that was, uh, you know, people continue to see value in having a knowledge base that's very accurate, and increasingly over the past few years, it's actually been even become more valuable because you need really great context for AI agents and assistants to work.

And I've got plenty of examples we could talk about that. So that's where that that response came from, uh, really, instinctively, is that, you know, I had sort of been through it in a couple different dimensions, uh, uh, prior to that. And just in terms of how I communicated to the team, the memo was actually like a series of memos. I mean, uh, in terms of every Friday, I send a company email. I just sent one right before I got on here.

And, uh, I am pretty transparent in that, in that, here's what's on my mind, here's what we should be doing, here's, you know, what, some great things that happened. Here's some people that demonstrate our core values. So I I've done that religiously for, you know, I've been at the company for six years. I do that every Friday. So the team basically knows what's on my mind. And so, uh, it was, you know, it wasn't like this big one big memo, you know, kind of tactic.

It was a series of emails leading up to this moment saying, here's what we're gonna, we gotta respond to this, here's what we're thinking about now, and so on and so forth, until it ba- basically, uh, I could put the flagpole down and say, hey, by the We Are Developers Conference, we have to produce a meaningful response, both on the public platform as well as on our enterprise front, because obviously there's great opportunity now to integrate AI into our SaaS application, uh, because that obviously is a different vector also.

So hopefully that helps.

Did you actually type the words code red?

Um, an equivalent, uh, I think I I've definitely, uh, I used, uh, disruptive, I used, you know, you know, existential moment, I used all those things, but, uh, I would I don't know if I use exactly the words code red.

I just think about that moment where like, all right, I'm gonna hit the C and the O, like I'm saying these words, it's happening.

Yeah, yeah. We have a very specific communication cadence with the community, with the company, obviously, like others. And the tone and the seriousness of what we were working on was very obvious to people, especially when you carve out resources and you take people away from certain teams, people are gonna ask like, wow, like, what about my stuff? And here you go, this is the reason, right? So it becomes very obvious.

How did you make those decisions to to pull people away? How did you decide which people? How did you decide which teams?

Certainly, I think with this is a hard problem to solve. So you certainly want, um, I think, very talented people, but I think certain types of people who are willing to break glass or go against the grain and not be sort of encumbered, uh, based on historical sort of norms. And so I think we I so very specifically picked, uh, a combination of people, uh, the people who were leading it were more newer people who had come from the outside of the company.

Because, you know, remember, we were going through a transformation. I joined a company that was engineering-led in 2019, all about this public platform, and we were transforming into this, you know, this, uh, product-led organization. So we brought I we appointed somebody that was, um, you know, very specifically, uh, if a kind of a newer person who had come from the outside, uh, and who was interested in building highly innovative, fast-iterating sort of products and had that sort of DNA and had that sort of, uh, drive to do it.

Uh, I also personally stayed much closer to it, um, and, um, you know, over I actually, in fact, ran product for an interim period of time myself, uh, with that person reporting directly into me. Uh, and so that was another way to sort of stay very, very close to what was happening on the ground until the the actual, uh, launch. And the rest of the team was a combination of very talented engineers, you know, designers, uh, and some people that had context of how the site worked in the past, who could, you know, provide us with all the, uh, the unlocks that we needed.

I think about Stack Overflow in in probably too reductive of terms in this context. You have inputs, you have outputs, right? The inputs are users answering questions, the outputs are the answers to those questions when people come and search for them. There's a whole community that makes that system run. The software platform sort of manages that community, and then you've got moderators, but it's really inputs and outputs, right? There's people who are asking questions and people who are answering questions.

Both sides of that are deeply affected by AI, right? And I think this comes to the open web part of the conversation, where the input side is being flooded by AI generated slop. And in 2022, you had to ban AI generated answers in Stack Overflow. And then the output side, you're the ability for AI tools to just supply the answers is overwhelming. So let's just break it into two parts. How did you think about the input side, where there's gonna be a flood of people saying, Oh, I can answer these questions faster than ever by just asking ChatGPT and pasting the answer in, and maybe that's not good enough, but I can just do it.

And then how did you think about the output side?

We noticed two things right out of the gate. One was the number of questions that were being asked and answered on Stack went through the roof. Because people started using, to your point, ChatGPT to answer these questions, and then they were able to, that just sort of fueled this kind of spike, which is kind of counterintuitive. But I think people just felt like, hey, wow, I can game the system, so let me just go do it.

And very quickly, uh, we are we are extremely shrewd and, um, you know, decide, you know, uh, our community members are amazing at sort of figuring out, figuring out what's real and what's not. And they were able to call out, uh, very quickly, that these, uh, posts were actually ChatGPT generated. And that's kind of what initiated the ban, which we completely supported and still support, by the way. So you still cannot answer any of the questions on Stack Overflow with AI generated content.

And the reason for that, Nilay, is because, you know, what we have is effectively, you know, our proposition is to be the trusted vital source for technologies. That's our vision for the company. Uh, you know, so for us, it's all about making sure that there are only a few places where you can go, where you're not dealing with AI slop, where you can actually, uh, with a community of experts have actually voted up and curated this in a way that you can trust, uh, for various purposes.

So, uh, on the input side, that it it made sense to do that, and we continue to do that. Uh, fast forward a little bit now, I would just say, you know, we have done many, many things to, you know, even though we've had high standards to ask a question on Stack Overflow, now we've created a whole sort all sorts of new entry points into the site. Our AI assist feature that we just launched actually in GA earlier this week, which has been super, uh, exciting to watch how users are using that, which is effectively an AI nav, uh, you know, um, conversational interface on our, uh, grounded on our 90 million questions and answers.

Uh, and then the ability for people to ask subjective questions. Going back to our last conversation three years ago, uh, now people are able to ask just open-ended questions. And it because there's a place for QA, which is the canonical answer to a question and so on, but there's also a place for discussion and so and and this conversation, because there's so much changing. So it it's not like all the answers have been figured out. So let's actually just make sure that people have an ability to do that.

And that's aligned with our mission of cultivating community, which is one of the three parts of our mission. The other one being power learning and unlocking growth. And so, uh, we have done all these things to make sure that, uh, the, uh, we're not restrictive on the entry point and the kind of the question asking experience. Um, the other, uh, thing that on the answer side, we also real realized that it's very important to go wherever the user is spending time.

So now that the world has changed and people are in fact using cursor and GitHub Copilot and any anything else to write their code, again, our goal is to be the vital source of intelligence. So let's show up wherever our users are. So we've actually become a lot more headless more recently. We've launched, for example, MCP servers for both our public platform as well as our enterprise product. And so what people are using our platform for to do now is not only, uh, invoke those MCP servers, let's say from a cursor when they're writing code and say, you know, how?

What's the difference between version one and version two? But also to be able to write back, which is very unique in the industry, to write back to our platform straight from cursor. If, uh, you know, they want to sort of engage on, sort of getting a deeper answer and so on. And so that's been our product principle, just go anywhere where the user is, uh, and hence, but ultimately, we just want to be the source, whether it's inside companies or outside companies, uh, to be that trustworthy vital source for technologists.

How do you monetize in a world where you're headless, right, where you're just another database that someone's querying from? Cursor, how do you? How does that make you money?

So our enterprise business, as I mentioned, is our s what we call Stack Internal, which is now used by 25,000 companies around the world, some of the world's largest, you know, organizations, banks, tech companies, retail companies use this product to be able to share knowledge internally. And now increasingly, uh, a they're able to use that trustworthy knowledge to power their AI assistants and AI agents to go do various things. A good example of this is Uber, uh, who has something called Uber Genie that is a customer of ours and with Stack Overflow internal.

They have thousands of questions and answers on our platform. Uber Genie plugs into that content through, uh, our APIs, and then it's able to go into things like Slack channels and automatically answer questions and drive productivity that way. So you're not bothering people, right? So it's it's try it's rooted in the context that's in the organization's, uh, knowledge, uh, on our platform. That's our primary business, the enterprise business. The second business, which we actually built only over the past couple years, is our data licensing business.

So one of the things that we also noticed, uh, was that a lot of the companies, uh, that, you know, the AI labs were obviously leveraging our data for LLM pre-training and post-training needs and dragon and RAG and indexing. We put up a whole bunch of anti-scrapers. We worked with, you know, third-party companies, and we did that, and very quickly we got calls from a lot of them saying, Hey, we need access to your data. Let's, you know, let's work together to, you know, formally get access.

And so we've had to do that, and now we've struck effectively partnership agreements with every single AI lab that you can think of for the for the most part, a cloud hyperscaler that you can think of, you know, companies like Google, OpenAI, all these folks, and, uh, and even partnerships with a long tail of, you know, the Databricks, Snowflakes, you know, the the some of the more, uh, kind of on the secondary, uh, point, even though they're not doing, uh, LLM pre-training.

And that's been our second business, um, more recently, and the third one, which is the smallest part of our company, is advertising. So, uh, I think most people assume that Stack Overflow is is supported entirely by advertising, but it's only about 20% of our company's revenues. And we have, again, a very captive, very, um, important audience of developers who do spend time on the site. And so we have, you know, large ad advertisers that want to get their attention on various products.

And in fact, now is a time when there's a lot of competition, so they they want to do that increasingly. So those that's how we make money. So in the context of becoming headless, uh, for us, it's about our enterprise product is, you know, it just works on a subscription and and kind of hybrid pricing. So that's how we make money there. The data licensing is similar in that they, if they want people on access, they got to pay for that.

And then, yes, advertising is limited to some of the largest companies, and, you know, they they pay us for that. But there's always going to be, I would say, it's an and versus an or, right? People are not, we're not going to be completely headless. I think we just want to give the user the option to be headless. Plenty of people still come to the site, and so in that case, uh, we're able to balance that out.

We're gonna take a short break. We'll be right back.

Support for the show comes from Upwork. Every business owner hits a point where they need more expertise than they can handle alone. But another full-time hire isn't always the answer. That's where Upwork comes in. With Upwork, you can hire highly skilled freelancers across a variety of specialties, from software development and AI implementation to marketing, design and business operations. You can browse profiles, review past work, and get help scoping the role. That way, you can hire with confidence and get started quickly.

And with Business Plus, you can access the top 1% of talent on the platform. Thousands of growing businesses already trust Upwork to hire highly skilled freelance professionals. It's free to sign up, and posting a job is simple. Visit Upwork dot com right now and post your job for free. That's Upwork dot com to connect with top talent ready to help your business grow, spelled U P W O R K dot com Upwork dot com.

This is advertiser content from Comcast Business.

Hi everybody, welcome back to another edition of Decoder Sessions. This segment features Andrew Melnitsky, the Verge's general manager, and Ivan Shefrin, the executive director for managed security services at Comcast Business. I hope you enjoy this conversation.

Let's talk about artificial intelligence. I want to understand how has AI, um, you know, changed the cybersecurity landscape for you and your team at the attack level.

The reality is it has had a dramatic impact on everything we do. The bad guys, uh, leverage, uh, leverage AI to for malicious and bad purposes. So the only way to, uh, uh, fight fire is with fire, and we so we now use AI just as the attackers are using it to break in more quickly, we have to pivot and use it to defend more quickly at scale.

Comcast Business specifically detected 25.4 billion phishing attempts and 21.9 billion drive-by compromise events over one year. Uh, talk to me a little bit about that data.

Phishing is so popular because it's so easy to do, and AI has accelerated that. Every business should have, uh, software in place to detect phishing attacks, uh, to train their users, and also that now extends to collaboration software, not just email, such as Zoom, such as Slack, such as Teams, um, because that's becoming increasingly prevalent. Everybody expects to get breached now, and that is not the attitude we use at Comcast. We we are take protecting our customer data extremely seriously.

Support for the show comes from Granola. Have you ever left a meeting thinking that was productive, and then ten minutes later, when it's time to take next steps, you're at a loss. Or worse, you've been heads down the entire meeting, furiously typing notes instead of actually listening, and now your brain is fried. Meetings are a mess. Granola fixes that. Granola is an AI-powered notepad that makes your meeting notes useful. It capture what happens in your meetings and turns it into clean, structured notes with decisions and action items pulled out and made easy to find.

And the best part, Granola works through your device's audio, which means it integrates seamlessly into the video conferencing tools you already use. No setup, no awkward bots. It's just your normal meeting with superpowers. You get to actually listen instead of frantically typing every word, and still walk away knowing exactly what was decided, who's doing what, and what comes next. Once you try it on your first meeting, it's hard to go without. You could try it totally free.

Just head to Granola dot ai slash decoder. That's Granola dot ai slash decoder to get your time back. Go to Granola dot ai slash decoder and try it on your next call.

Black Friday is around the corner, your biggest chance to grow sales this year. Don't wait. With Realize, Taboola's advertising platform for businesses, you can reach people engaging with your competitors across premium websites with 600 million daily active users. And as an exclusive offer for podcast listeners, new customers get $700 credit back. Just search Realize Podcast on Google or use the promo code Podcast twenty six. Ready to get started? Take advantage of this offer while it lasts.

Terms apply.

Welcome back. I'm talking with Stack Overflow CEO Prashanth Chandrasekar about what an existential crisis for the company the launch of ChatGPT was. And of course, that led me to ask, why would anyone new come to Stack Overflow now? Do you think new users are going to come to Stack Overflow? Stack Overflow is a product of the mobile era, right? There's an explosion of software development. There's an explosion of community. There's a culture and the value of building apps and services, and there's new tools.

Stack Overflow is one of the central points of gathering for that community in that era. New developers today might just open Claude Code or Cursor or GitHub or whatever and just talk to that and never actually venture out into a community in that way. Do you think you can get people to come to Stack Overflow directly and seek out answers from other people, or are they just going to talk to the AIs?

I think that the the for simple questions, uh, by the way, when we saw the decline in questions early on in 2023, we what we realized is that pretty much all the question declines had to do with very simple questions. Complex questions absolutely get still still get asked on on Stack because there's no other place. You know, the LLMs are only as good as the data, which is typically human curated, and we're one of the best places for that, if not the best, uh, for technology.

It's still a very active site, a lot of engagement, a lot of monthly active usage, and the questions being asked are quite, um, you know, I would say they're advanced questions, but what we're also increasingly seeing through our new mechanisms that we've opened up, because we want to answer to answer your question. We want to give people other reasons to come to the site other than just getting their answer. So we have had to broaden our site's purpose, and hence the mission of cultivate community, power learning, and unlock growth.

So what we've done with those things is let's open up new entry points, new ways for people to engage. So we've in fact, for example, Nilay unlocked the ability for humans to chat with each other to get directional guidance. So that's been a very, very popular feature on the site where people are, uh, engaging with other experts. So we have, for example, people asking OpenAI API questions as an example, and they can go into the OpenAI chat room and be able to engage with other people having, you know, uh, that have similar questions as an example, or Python experts as an example.

We also opened up ability for people to get, um, uh, demonstrate their knowledge by opening up challenges. So we, you know, effectively like hackathons. That's another idea that we've got where we've opened up a whole bunch of series of challenges, very popular feature. Now people spend time to go and solve these challenges that we post, and then that way it can accrue to their ability to showcase their understanding of the fundamentals, which I think is very important in terms of the world where the world is going.

Because if people are just using vibe coding tools and, and, uh, you know, co-gen tools, I think companies that are bringing in young talent ultimately have to know that they're relying on the people who not only took the shortcut, but also understand the fundamentals. And so we're one of the few places where you can actually prove that you've actually learned the fundamentals, uh, you know, so that's that's that's the other reason why we've opened up these new mechanisms.

And then the third reason, uh, third part of the mission, which is unlocking growth, is we also want to enable people. There's going to be a lot of job disruption, uh, as a function of all this, and people's jobs are going to change quite dramatically. Junior developer jobs, even though I think it's a shortsighted move by many companies to stop hiring them, considering you need a pipeline, those people are going to need a home. Then you're going to need to connect with other people to be able to sort of progress and learn and get get jobs.

And so job is a very important part. We struck a partnership with, uh, with Indeed this past year, so that they, you know, be, uh, you know, partner on tech jobs. So it's just to broaden the scope of our site, so that there are many other reasons other than asking the question, which they will still do, but we want to give them more reasons to come to the site also.

This comes, I think, to the the big tension in all of this. And I I see it playing out on all kinds of different communities. I see it playing out in our own comments in a lot of ways. You want to build a community of people who are helping other people get better, and that is being disrupted on every side by AI. And communities that are built around people are pretty resistant to the incursion of AI.

oh, definitely.

This has definitely happened on Stack Overflow. Your moderators have essentially revolted over the ability to remove AI generated answers as fast as they want to. When you partnered with OpenAI, a bunch of users started deleting content so it wouldn't be fed into OpenAI for training, and you had to ban a bunch of them. How are you managing that balance? Because if you build communities around people, I would say right now anyway, the the culture is those communities will push back against AI very hard.

It's it's I would say one of the most important things that we've focused on, and I've spent time on over the past few years, is this whole push and pull, as you describe it, of like, should we? How do we think about AI in the context of our site? You know, because we're it's pretty clear to us and to me that, you know, if we don't modernize the site in the context of, uh, us leveraging AI as an entry point, et cetera, that it's going to be less relevant over time.

And so that's that's not, that's not good. So we've taken a very aggressive stance by incorporating AI into the public platform now with AI assist, as I mentioned, which has been fantastic to see. And I'll walk you through the decision on why we did that, and then same thing on the enterprise side. Um, there's definitely, you know, if I think about the user base at Stack Overflow, it's kind of like a big nation, right? Like we've got 100 million people, and there's definitely, uh, you know, people on both sides of the spectrum.

What's interesting is that there is a, you know, we have something called the one nine nine zero rule. 1% are the hardcore users who have been spending a lot of time, you know, with their blood, sweat and tears curating knowledge, spending their time on the site, et cetera, contributing 9% doing it in a medium way, and then 90% sort of consuming and mostly being a lurker on the site. And the the distribution of the site, I mean, we ask people on whether or not they're using AI.

Our own surveys, basically, say, if you took a look at the Stack Overflow 2025 survey, over 80% of our community members are using AI or intend to use AI. 80%, right? But, you know, the the trust level on that answer on on when they're using AI is only about 29%. Only 29% of our user base actually trusts what's coming out of AI, which is actually quite appropriate considering where we are, because, you know, there's there should be skepticism with this new technology.

So there's enthusiasm to try it, but not to fully trust it. And with this one nine nine zero rule, I think what we have is we have a core group of users that are always going to be the protectors of the original mission of the company, which was to create this kind of knowledge base that was completely accurate and do nothing more than just that. And then you're going to have we have a very large number of people who are, let's say, younger developers, people, the next generation of developers who are looking to leverage the latest and greatest of tools.

And it's very clear to us, based on surveys that we've done and additional research, that they want to use natural language as the interface to be able to do this. It is the most meaningful sea change sort of change in terms of, you know, computer science development. You know, if you look back to all the way, you know, to op, you know, even, uh, object oriented programming many, many decades ago, uh, that wasn't a such a huge boom, actually, like it didn't actually create this sort of change.

But now we're in this moment where everything's been unlocked. So I think it's it's a huge change effort, and we've had to sort of decide that, hey, we've got to be able to respect the original mission, keep accuracy at the heart of it. Um, and so we're not comfortable with using, for example, AI for answers, because it'll generate slop, because it hallucinates, and that's hence the trust score is low. But why don't we incorporate natural language interfaces?

So that's the preferred way to sort of engage, and so what? So we ended up doing that both on the public side as well as on the enterprise side of the house. Uh, and that's been really well received by the vast majority of users. But there will be always a vocal minority that will push back against incorporation. Because there's a lot of, you know, beyond just the site, there's just a a level of, I think, broader concern about what all this does to jobs.

And if we let the, you know, cat out of the bag, then what's going to happen, you know? So I think there's there's that obviously concern also, which is understandable.

Let me put a pretty fine point on that. I I think I understand that in a maybe a sharper way. If I am somebody who in in in your 1% who spent a lot of time on Stack Overflow helping other people, and the reason I answer questions for free on your platform, which you monetize in lots of ways, is because I can directly see that my effort helps other people grow. And I'm helping other people solve problems.

That is one very self-contained dynamic. The last time you were on the show, our entire conversation was about that dynamic, right? And how you and how you got people to participate in that dynamic and the value of that dynamic. And then suddenly, there's a very clear economic benefit to the company that owns the database because they're selling my effort to OpenAI, right? Which is a thing that is happening across the board, right? We're gonna, we're gonna do all these data licensing deals with all these AI providers.

They're gonna train on the answers that I have painstakingly entered into this database to help other people, and now the next generation of software engineers is gonna get autocomplete that's based on my work, and I've gotten nothing. I mean, I've heard that from lots and lots of people, right? I've heard that in our own community, I think I have felt that as various media companies have made these deals. How do you respond to that? Because that feels like what you were providing was a database you had to monetize in some ways, but the people, the interaction that the people had was the value.

And now there's another kind of economic value that is maybe overshadowing it or recasting or recharacterizing the interaction that the people have.

there are a couple of points there. One is, I think if you think about the original DNA of this company and why people came together to do this thing, you know, when I joined the company, I asked a question like, what's people's incentive to spend time doing this? And I asked the founders, uh, specifically Joel Spolsky, about this, and his point is that the software development community is effectively it's a very altruistic. People just want to help each other out, you know, because people understand how frustrating I I used to write code many years ago.

I recently picked it back up with, you know, some of the code gen tools, which is interesting, you know, compare and contrast. But I just remember how frustrating it was if you get if you get stuck on something. And so Stack was obviously a huge boon, uh, when it was created, uh, to unlock this. And so it was truly out of that. That was the reason. I also we also asked the question like, should we along even before ChatGPT, should we incentivize users by paying them for that?

Should we g give them a monetary benefit? And that wasn't like a high ask by our user base. We went and researched people. So people were not in for the money. Plus it it complicates things because how do you judge the, uh, the payment for a particular JavaScript question relative to a particular Python question? You know, it's just a very, very it it goes down a rabbit hole, which is, I think, uh, a kind of a sort of a untenable sort of rabbit hole.

And so this commercial aspect, so that's one. So what is the original, you know, reason why people got together and it was about the mission. Secondly, um, in terms of like, I think the why we have to do this, and you know, is it, uh, is it unfair? And so on. The primary reason why we have to go down this data licensing route or why we've had to do it, um, is because the model of the internet has literally been, you know, has been turned upside down.

People relied on, and I know you talk about this, uh, Nilay, with the the DoorDash problem. I think it it the model of the internet has literally where people are, you know, go to search engines and go to, you know, websites, and, you know, uh, you monetize off of ads. You know, that is completely I I really empathize with, you know, content sites that are heavily dependent on advertising. Because I think most, uh, content sites, their traffic is down thirty to 40%, something like that.

So with this huge sea shift, and where companies that support these platforms have to ultimately their wear a business ultimately. So what do we have to do? We have to do what is necessary to adopt a new business model to survive and thrive and do all the things. And so thankfully for us, we've had an enterprise business, which is independent of all of this. Thankfully for us, we were now able to, uh, we still had the advertising business for large advertisers still cared about this, uh, you know, on our community.

And so data licensing only felt right in terms of making sure that we can, lever, you know, effectively, uh, capitalize on the moment, plus also, uh, be able to invest back into our community so that people who are there for the right reasons saw the benefits of that. So we've invested with all these new features I just mentioned, you know, whether that is all these new content types or challenges or chat or AISS, or any of these things, all it takes resources to go and build.

And so we've had to go and, uh, and we leveraged these funds that we received, uh, to be able to go do that. Now, in the future, we may consider other ways to make it, uh, uh, you know, um, for example, should we pay our users, give them a piece of, uh, piece of the data licensing revenues? Perhaps, you know, we we will ask that question always. There's always ways for us to, um, continue, but not right now.

This this is the current setup that we have.

You mentioned to to get to the data licensing deals, you had to put up a bunch of anti-scraper tools. You had to go, you know, into secondary and tertiary layers of the stack to get deals from Databricks and other kinds of providers. The AI companies, they were just scraping your site before, right? They were, I mean, they they probably still are. Like whether or not they're paying you, they're they're probably still just going through the front door, because that all of them appear to be doing that.

Did you have to say we're stopping you and then go get the deal, or did you say, hey, we know you're doing this, but you have to pay us or we're going to start litigating?

Somewhere in between, I would say. I think we we put up, we put up the anti-scrapers very quickly. Uh, we even sort of changed the way in which people, uh, received our data dumps. And, you know, because we want, again, there's a balance, because we never wanted to prevent our community users from grabbing our data for their legitimate needs, you know, for them to do their school projects or PhD theses or anything like that. So we've continued to be open about our data for our community members, but they have to be community members, and they it can't be companies looking to commercialize off the data.

So we were very specific about the policy terms, um, uh, putting up technology that prevented people from grabbing it. So we know exactly who is scraping, who is not scraping. Uh, we some parts of those were outreach to those folks to say, look, you know, stand down because you're just you're you're obviously putting up even a pressure, you're putting a lot of pressure on our servers by doing what you're doing. So, you know, uh, take it easy, or, and here's

But but I I I think my characterization of those companies is they don't care, or some of them care and they they want to be good citizens, and some of them absolutely do not care and they would prefer the smoke. You can just categorize them. There's a reason Amazon is suing Perplexity, like they've told Perplexity to stop it, and Perplexity won't. As we're speaking today, this morning, the New York Times is suing Perplexity. Then there are other players who are acting in different ways and they're striking different kinds of deals.

Walk me through one of those deals. When you went and struck your deal with OpenAI, was it we're gonna stop you? And if you want the door to be opening, you have to pay us, or was it, you know, this is wrong. We can take all the technical and legal measures, but we should actually just get to the deal correctly. Like, walk me through that conversation.

with some folks like them, we were already, you know, we were incorporating something like OpenAI into our product. Remember the code red situation where we were about to announce our AI response? So we were actually using their technology to do what we had to do, uh, to incorporate AI into the public platform as well as our enterprise product. So we had a relationship with them, and we also said, Look, this is not gonna work, it's not tenable, and this is the new way of working.

And so meaning we need a new arrangement, a business arrangement for you to use the data. And so let's actually have a conversation. And to credit to them, they were very partner-centric around that. I was very impressed by both OpenAI and, you know, companies like Google, who are all very open to engaging on this topic, um, and wanted to be responsible AI partners. They just they got it immediately, even before we asked them. It wasn't like this kind of big, you know, let's go have this conversation kind of from the ground up and justify why it had to be done.

We just said, Look, this is what needs to happen because it's a new business model. And we got into the conversation pretty quickly in that, okay, let's actually have a constructive way to, you know, what exactly are you looking for? Which format of data do you want to scrape? The content? Do you want bulk uploads? Do you want API calls? You know, what do you want? So we got into that whole mix. And then, of course, there is the conversation around, uh, and these are recurring, just mind you, uh, Nilay, that these are recurring revenue type of deals, so these are not one time, um, you know, um, payments.

Uh, so, yes, it was, uh, they were very great, very collaborative partners. And but you're right, there are players who, uh, uh, you know, uh, are they're they're contradictory. They say something and their their actions, I think, prove others other things, uh, you know, in terms of how they've engaged. And so there are people that are holdouts for sure, uh, people that are not exactly consistent with their word. And uh, that's unfortunate, you know, it's it's a an and I think every company like us has to decide what we do about that.

And, you know, we're in various stages of these conversations with various people on, you know, how to make sure that we, uh, ostensibly get them to sort of do the right thing.

Now you have to name one of those companies. Who who do you think is is holding out differently than their public posture?

I'd rather not be specific, but I think you've actually, like, characterized all the usual suspects that you are you that you are covering are the usual suspects that, you know, we are encountering, is how I would put that. Yes,

Let me ask you about the recurring revenue piece, and then I want to get into the decoder questions because I I I think they'll be illuminating after this conversation. There's a sense, right, that we've done all the pre-training that we're going to do, right, that scraping the internet is not the future of these models, that there's there needs to be some other leap. Stack Overflow's existing corpus of information is the valuable thing, right? It's the it's there's a lot of information there.

It's twenty years of of stuff in that database. What's the value of you have to pay us again to train the next version of Gemini or GPT? And the value of there's incremental information being added to the existing database. Because that seems like a clear split to me.

the way that we have thought about this is that, you know, every model that's being trained, you're training it on some corpus of information. You know, you're going from GPT X to Y, and so in the new model that you're training, you if you're leveraging our original data or some derivative of that from a prior model, then you have to pay us for it. That's effectively the legal requirement, uh, to be able to do that. And so it's it's a cumulative aspect, right?

So let's not forget that. And so, like, people have to pay for the cumulative data. It's not just the fact that it was used in back in the day. And yes, obviously, relative to twenty years, you know, one year's, uh, worth of information is going to be less. But, uh, but that's why you're getting twenty plus one, you know, that's the idea. Um, and so that's just the way the the legal agreement has been set up.

Is it per year? Is it every year's worth of data is a a chunk of money, or how does that work?

Uh, no, it's just it's a cumulative, uh, it's like the whole corpus past, you know, historical data, as well as, uh, you know, the the, uh, anything that's that's going forward for the following year, um, all that is sort of one accumulated sort of data set, and that's, you know, charges one, effectively.

But that doesn't so, like, this year's data doesn't get pulled into the training set for Gemini three, which just came out, right? Every new question answer in Stack Overflow since Gemini three has come out is not incorporated in Gemini three's training. So you're kind of betting that they're just going to train ever bigger models. Is that is that how it's structured in your your mind?

That's right, that's right. Yeah, yeah, and it and some companies have asked, you know, they've asked for, you know, very, very w- they're wide, uh, use cases, you know, there's, um, pre-training use cases, there are, you know, uh, even beyond that, you know, you can leverage the data in many, many different ways for AI and non-AI use cases, uh, search use cases, and so on. But correct, I think that we in some they may be scenarios where more large larger models are built, and our data is going to be useful for those scenarios.

But there's going to be RAG indexing, there's going to be post-training needs, there's going to be all sorts of scenarios, and it's quite interesting to see some of the frontier labs ask for very specific slices of data that they find it to be, you know, find very useful. And we're able to remember, we've got a lot of there's not only just the question and answer, but we've got the comment history, we've got the metadata history, we've got, you know, the voting history, we've got the user A has gone down this path history, uh, you know, so it's it's a lot of excellent context for things like reasoning, um, and to really sort of be able to be useful to really

mimic the the human brain. It's effectively one human brain that's been documented, almost.

We have to pause here for another quick break. We'll be back in just a minute.

Support for the show comes from CrowdStrike. AI is already fundamentally changing how companies build products and run their businesses. If you listen to the show, you probably already know this. But while AI drives innovation, every new model and agent also creates new cybersecurity vulnerabilities. That's why CrowdStrike built AI Detection and Response, or AIDR. AIDR helps organizations discover where AI is being used, monitors AI activity at runtime, and helps teams detect, investigate and respond to threats involving AI systems.

As companies adopt AI, security has to be built in from the start. More than 70% of the Fortune one hundred trust CrowdStrike to protect their businesses. Now, CrowdStrike is helping secure the AI era. You can learn more about AI detection and response at CrowdStrike dot com slash decoder.

Support for the show comes from Framer. If you want to create a website that looks great, feels handcrafted and comes together fast and seamlessly, then look no further than Framer. It's a complete website platform, not just a builder. So teams can launch and keep improving their sites in one place. It helps creators, teams and businesses ship production-ready sites faster than ever while getting every detail right. You're able to prompt, inspect, edit, and publish in one place at a whole new pace.

On Framer, agents and humans work in tandem. Agents bring speed and scale, while people bring taste, judgment and control. And that all takes place in the same workspace where the real site is designed, managed, reviewed, and published. Thousands of businesses, from early-stage startups to Fortune five hundreds, are choosing to build their websites in Framer, where changes take minutes instead of days. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer dot com slash decoder for 30% off a Framer Pro annual plan.

That's framer dot com slash decoder. Rules and restrictions may apply.

As soon as autumn rolls around and you swap your iced lattes for pumpkin spice, that's your signal to start thinking about the holidays, the decor, the gifts, the shopping lists, but more importantly, the holiday treats, cheesecakes, cookies. Is there anything better? And when the recipe calls for cream cheese, there's only one option Philadelphia, the original cream cheese, made with fresh milk and real cream.

This holiday season, whether you're going all out or keeping it simple, make it really, Philly good. Find your favorite holiday recipe at creamcheese dot com.

Running a business shouldn't feel like surviving a software group project. One app for accounting, another for inventory, another for sales, and somehow none of them talk to each other. That's where Odo comes in, an all-in-one business management software that brings every part of your business together. From sales and accounting to inventory and marketing, all in one powerful platform. No messy integrations, no bouncing between tabs, and best of all, no spreadsheets. Stop managing software and start managing your business with one unified system.

Try for free today at odo dot com slash vox. That's o d o dot com slash vox. Welcome back. I'm talking with Prashant Chandrasekar, CEO of Stack Overflow. We've talked a lot about AI and models and disruption. These are all big issues. So how do they fit into the fundamental work of being a CEO, which is to make decisions and keep the company running? This, I think, brings me to the decoder questions. Uh, you've restructured the company, there's been some rounds of layoffs, you've obviously refocused on the SaaS business in a real way.

I think we should talk about that. But the idea that we're going to train ever bigger models and that will be the growing part of the business versus we might just want some slices versus it's actually retrieval augmented generation or RAG that's going to be the future for a lot of these other businesses. You would make different decisions based on which one of those is going to grow faster. And I don't think anybody knows. Correct, yeah, May maybe you know, you can tell me if if you you know, or you know someone who knows, but we're in a very nascent period for all of this kind of development.

How have you structured the company to even out all of that risk and and be prepared for how do people actually need the data in the future? Because I don't, maybe you know, but I don't think anybody, I don't think I know.

yeah, yeah, it's it's hard to predict clearly, uh, and also, you know, who knows, uh, with, you know, we've got some brilliant minds like the the Demis, uh, Hassabis at Google and others who are coming up with the next generational leap of whatever the equivalent of transformer technology is to, you know, to to keep going towards this ultimate goal that they're all pursuing, which is AGI. Um, so, yes, you're right, it's hard to know exactly what shows up, um, and when.

However, the way we're structured as a company is effectively two parts, Anile. So the one part is the enterprise business. So we have a product team, engineering team, we have obviously go-to-market, uh, go-to-market team focused on that, and that's very, very clear, the enterprise products business. And then the other side of the house is what we call the community products side of the house. And the community products team focuses on the public platform, all the features that we talked about so far, uh, AI assist and all the subjective new questions and chat and all these things, uh, and this the community site and the data licensing business sits in that group.

And so that they are obviously tied to the, you know, the kind of the engagement on the site. And so there's a kind of a a a virtuous cycle there. So that's how we're split, and that includes again, product resources, uh, you know, a small go-to-market team, engineering folks, et cetera, et cetera. So that's, um, and then of also a community management team, which spends a lot of time with the moderators, et cetera, uh, to engage there.

So that's so it's split down the middle, and of course, our our other functions support both.

How big is Stack Overflow today? I know you you laid off almost a quarter of the company in 2023 because of traffic declines, you've built other businesses. How big are you today?

We are about 300 people or so.

And do you think the revenue you're going to see from data licensing or your SaaS business is going to allow you to grow again?

We believe so, yeah, we're a growing company, we're profitable, so, you know, financially, we're thankfully in a very good spot. Uh, and now it's all about placing bets on the, uh, on the highest growth opportunities, uh, and we believe creating this knowledge intelligence layer inside the enterprise through our Stack Internal product is is a phenomenal growth option because customers are pulling us in that direction, which is fantastic to see. So,

That's where you're headed. Those are so many announcements. I do want to wrap up by talking about those. I I'm just focused on the future is the SaaS business for enterprise, the future is data licensing. Are you still seeing declines on the public website?

I would say it's been stabilized, uh, for a few months. So I think the engagement and the activity on the site are actually pretty stable. The the the drop in questions that I was mentioning previously were all the simple questions, and it seems to be now has sort of come to a place where it is, uh, you know, the complex questions are being asked. We have a consistent number of people on the site every day. We have something called a heartbeat.

In fact, anybody can go check it out. There's always, uh, you know, if you go to StackOverflow. com, you'll see at the bottom how many users are online at the moment, sort of thing. And so you'll always see, you know, a very consistent number there. So that's, uh, so it's I think it's, uh, it's hard to know how to predict the future, but certainly I think the the the worst of it was back in, you know, twenty three, twenty four, for sure.

The question I ask everybody on Decoder, as you well know, is is how do you make decisions? Last time you were on the show, you said that you wanted to be on the front lines as much as possible, uh, and you wanted to be informed by people who are on the ground. Has that changed in the past few years? What's your decision-making process?

Yeah, not not really. I think it's, uh, very important, uh, for leaders and, uh, people like, uh, you know, CEOs, et cetera, to have the full context because these decisions, uh, cannot be deleted you know, kind of, um, uh, you can't have filtered information. So I spend a lot of time with users, a lot of time with customers, uh, really understanding, uh, what they care about, and that's how we've even decided, even something like as controversial as the AI assist feature to our productization.

It it wasn't obvious if you were not, you know, on the ground, uh, to say, let's go and, um, you know, build that, because if you just listen to the headline statements, it seemed like, hey, people don't want that, wanting to create AI into Stack Overflow. But the reality is that many, many users, the 90% I was mentioning, uh, you know, wanted a natural language interface. That's what they are comfortable using these days, and that's what they wanted.

So that's why we decided to do that.

One of the things that I I see everywhere is that split you you mentioned it before, one nine ninety, right? There's a very vocal minority. We see it in our own traffic on the verge. We we cover AI deeply. We are told that everyone hates it. I understand why, I understand the I understand the comments. Uh, that's what I'll say. I get it. And then I see the usage numbers, I see the traffic on our coverage of AI tools.

I I see, eh, companies like yours saying everyone's using it, and there's some gigantic split there that is unlike any other split I I think I've ever encountered in covering technology for the past fifteen years, where everyone says they don't like it and then they're using the hell out of it. The only one other one I can think of that is lightly comparable is how people Feel about Adobe. Everyone uses the tools and everyone's mad at the Creative Cloud subscription fee.

Is basically the only comparison I have. It's not a good one, it doesn't map one to one, but it's as close as I've come to that split. What in your mind accounts for that split with AI, where people don't like it, they're very vocal about not liking it, and then we see the numbers and everyone's kinda using it anyway.

I think it comes down to that data point I shared earlier, which is that eighty plus percent of our user base wants to use AI or already using AI for code-related topics, but only 29% of that population trusts AI. And trust is a very deep word, you know, it's like, like, why why don't you trust something? You don't trust something because you don't think it's producing high-integrity answers, uh, accurate answers, uh, you may not trust it because it may replace you one day, uh, and, you know, you you you don't like that either, so but at the same time, you're you're obviously gonna be curious on what is this, uh, this force that's going to be such a

demonic force, uh, and so you wanna keep trying it and and using it, and be, uh, perhaps getting better and ultimately, hopefully, uh, leveraging it to your benefit in a way that you can be le- relevant as an individual developer in the future, uh, being going able to go a lot faster and so on. So, I think that's probably the c- pro probably the reason because I think, especially the developer audience, I think they're very discerning, um, a kind of a, let's say, analytical audience, and they can they can be prickly if the if things are not, you know, uh, let's say, deterministic, right, the way it has been for a very long time.

And so this is a very probabilistic sort of technology technology. So the fact that you it's almost like, you know, uh, going to a casino and using a roulette wheel, you're gonna get a different answer every time, uh, it's it's not necessarily comforting for somebody who's writing very, very specific, uh, code, looking for very specific outcomes. And I think that'll people will get used to that over time. It is a mind shift change for people writing software, um, and that may be the reason that why people are intrigued because it is so powerful as a technology.

Don't get me wrong, we we use vibe coding all over the place at Stack, right? All the features I mentioned to you, we our designers and our product managers vibe coded it first to show it and get, uh, user feedback before we went and built it. So we've embraced these tools internally for their benefit. So there will be ways in which you can you you feel comfortable using it, but still, um, I think that's that's the core reason.

I actually wanna talk about that dependency, right? You know that it's not trustworthy, but you are building products with it, you are building products to enable it. Th- you've mentioned it several times. The big announcement this week is Stack Overflow AI Assist. You've talked about it several times throughout this conversation. You're betting that this is what people want, right? You're betting that an AI-powered tool on Stack Overflow will help more people, and then maybe that thing is gonna hallucinate like crazy and give people the wrong answer.

How do you make that bet when you know that the users don't trust it, but you still have to roll out the tools cause that's where the industry's going?

We believe we've actually unlocked a very important aspect of, uh, you know, that trust issue, which is, uh, and responding to it by our AI assist a feature provides it's a RAG plus LLM solution. So effectively, it provides an answer that first goes out into our corpus of information of, you know, uh, tens of millions of questions and answers. We have eighty to 90 million, and that those are first used to produce a response. And then if they don't, then there's a fallback option where it goes and leverages OpenAI's, for example, who's our partner, to be able to go and produce, uh, uh, trustworthy knowledge from other parts of the of the of the web.

And so it's pro it's first searching through our trusted attributed knowledge base. It pro produces the links so people can go go down that path and learn more about it. Uh, and so we're th- which is very important to us, right? Attribution and and and so on. And so that's how we are navigating this element of hallucinations, and we're constantly testing it, and it's n- it's not perfect, right? Uh, there will always be improvements, but we are also looking at where the world's headed, and if these models continue to get better, then this should, uh, or we should benefit from those improvements, and ultimately, we sh- we should have the best solution because you've got grounded human context plus the

LLM, uh, uh, the LLM strands as well.

I think the thing that I'm most interested in is the faith that the models will continue to get better. I'm not 100% sure that's true. I'm not sure that LLM technology as a core technology can actually be intelligent. Uh, right as you're saying, people are very attracted to the natural language component of these models and the interface shift that's happening, the platform shift that we're all gonna develop software with natural language or let the LLMs reason, basically self-prompt themselves into an answer, right?

There's something there that seems risky. Are you perceiving that risk today? Are you factoring that in, or are you saying this is where we're at now, and we we have to continue until something changes?

I think the first thing around the improvement on the LLM, uh, let's call it plane, I'm with you, right? Like, it's hard to know how, uh, things are gonna improve when you when you just think that okay, it's plateaued. When you thought about the past six months, boom, here comes Gemini three, uh, and again, We're h- proud partners of Google, and here we go, you know, it's it is a seed shift. It blows every other model out of the water, and now we've got a code red situation in other companies, other LLMs, competition.

So, you that's every Sam Alman did Day

Type the words code red by the way. I Wanna be clear that So that's very good.

Perhaps, uh, perhaps that that was the way for me to go back in the day, but I would say, uh, but the the point being that that is that it's so it is it is surprising that you're able to produce that sort of a leap when seemingly things have plateaued. So I don't know, I'm not a I can't predict that, right? Uh, nearly cause, You know, these folks are deep, deep in the subject, uh, uh, Demis and others, and so th- that's true, but there's also gonna be other, I'm sure, innovations that we are not even privy to, uh, like I was explaining previously, like if, you know, uh, transformers were obviously a huge development in the space, they may

be something that, uh, these AI research labs come up with that we we're not even aware of. That's going to be ultimately pushing things. Ultimately, we know that the compounding effects are very real. You know, we've got unlimited compute, you've got, you know, w- extremely powerful chips and GPUs that are now even lowering their cost. And I was at, you know, AWS reInvent this week where they talked about the Trainium chips and Trainium three and Trainium four being built out.

So, you know, there's going to be just the proliferation of these compa- and then you got access to data, uh, that, you know, what we've already talked about, and so these things, when they combine and compound together, it's gonna produce very magical outcomes. So I think that's the belief and why it's rooted, uh, you know, why my why why my you know, my own assumptions are rooted in the fact that it's gonna improve.

The split, and the reason I'm asking about this in the context of the tools you're building and everyone using it, and only 29% of people trusting it, is you gotta bring that number way up to reach the returns that every company investing in AI is is trying to reach, including yours. I don't know if the core technology can do that. I don't know if you can stack a bunch of technologies to do that. But I do know that one version of the future of software development looks like everything is vibe coding all the time.

Another version of the future of software development looks like writing intensely long prompts For models that are pages and pages themselves, which seems ridiculous to me, but, you know, maybe that is the future. And another version looks like, oh, we return to humans at the cutting edge of software development, and they are co-developing with an AI model, and then maybe asking stack and like, that feels like a richer, more interesting future, but it's unclear to me where we are on that on that spectrum or how that e- even plays out.

We obviously have a, uh, not only a bird's-eye view in the context of our public community with this, you know, 29%, uh, data point as an example, but we our enterprise customers give us a clear view on where they are, right? Because the ROI question's being asked very heavily inside companies, and clearly, uh, you know, I think you the 2026 is gonna be the year of rationalization. If 2025 was the year of agents, at least where every s every tool's being tried out inside these companies, there's a very open, like, uh, open la- a landscape for CTOs to go and buy various tools, test various tools.

And so I think it's it's been a I think a tremendous time for some of the the the companies building these tools. Uh, but twenty six, the CFO pressure with, Hey, look, okay, these productivity improvements have to come from these. We're gonna hire less people as annual planning happens. You know, there's gonna be tremendous pressure on the system to prove out what the real the real value is. Everybody at at a senior level that I've talked to acknowledges that this is a this is a big shift, and they all are leaning into it pretty hard.

They're all waiting for the improvements. I think most people, most companies will say that they've seen improvements in the small groups where they've tested these tools, but that's a sort of a self-selecting group cause they're the enthusiasts and so on, and they will see great productivity gains, which is probably true. But there's a absolute drop off in productivity as you think about the the, uh, adoption across the enterprise, and probably for the same reasons, by the way, you know, you're telling employees to use tools which may put them out of a job.

So why would they wanna do it? Or more fundamentally, if these tools are not perfect and they're hallucinating and, you know, they they're gonna be held accountable, then that's not good either. So a and then, of course, is the process change, the mindset change, the the ability you have to completely change the workflows of how you work, all the enterprise change management work. Which is why our solution, which is Stack Internal, is building this human curation layer, a knowledge intelligence layer, so that you can, uh, with the MCP server on top of it, the knowledge base in the middle, and then our ability to ingest knowledge from other parts of the company to create these atomic Q and A that

are extremely helpful to root your enterprise knowledge in, uh, through these AI agents. That is that solution that we that's why we've gone and really gone hard at producing that. Um, and we've seen a really, really strong response from our customers. We have some of the world's largest companies, uh, you know, leveraging and testing and building this with us, you know, HP and Eli Lilly and all these companies, Yurts, um, and so it's been amazing to see them, uh, gravitate towards us cause they want to fulfill that ROI point that you're Making.

Which is and what are the gaps? It's trust again, and so they want this trust layer, uh, through a company like Stack that they can actually insert in between their data c- data as well as their AI tools.

When you say the age of rationalization, what I hear is you think the bubble's gonna pop in 2026.

Yeah, I think certainly the exuberance in just trying out various tools and, you know, uh, you know, unlimited budgets on this AI budget, I think will ultimately it'll come to roost. Um, I'm not sure about the bubble bursting, like, I I think that there's definitely gonna be corrections along the way, there's no question about it. That just like, you know, if you look at history, um, but the number of vendors that are selling into these companies, what I'm very surprised by is that there's similar functionality, there's four of them being tested within these companies.

You know, certainly, all these companies may have received what if they've got 100 million in ARR, but at some point, there's gonna be churn when the CTO decides, you know what, I'm only gonna use maybe one, uh, and maybe a second one as a backup. No different from the cloud, you know, if you think about the multi-cloud world back in the day, uh, you know, people didn't have three clouds out of the gates. I mean, now you have maybe one primary cloud and one secondary cloud.

But this is different, but at the same time, I don't think you're gonna have, like, four different vibe coding tools, in my opinion.

When you think about Stack Overflow as being that trust layer, right? I mean, that's the value add, that's what maybe you can charge a premium for it over time. You're still dependent on it with tw- only 29% of your users trust. How do you and I know you're talking about RAG and that your other systems are doing that. How do you think you bring that number up with Stack Overflow? Is is that possible for you to do, or is that the ecosystem has to do it for you?

I think it's an ecosystem point, uh, more generally speaking, because, you know, that is more a reflection of people's, uh, you know, what they have access to beyond stack, right? They've got access to all these other options. And so what we can focus on is being the most vital source for technologists. And so for us, it's about making sure this content is excellent, it's high quality, but also it's a great place for people to, again, cultivate community, but connect with each other, learn and, you know, grow in their careers.

Um, but the I think the the way in which we can do it is through the fact that we are working with all these big AI labs and the fact that our trustworthy knowledge that's been human curated painstakingly is going to flow into these, uh, LLMs, which ultimately produce trustworthy answers. So we're sort of one layer behind, but that's that's, I think, where we operate. We operate in that trust layer or the data layer, if you will, uh, in the context of LLMs.

So that's our indirect contribution to that 29%.

Prashant, this has been a great conversation. You're gonna have to come back sooner than three years next time. What's next for Stack Overflow? What should people be looking for?

Our biggest focus is going to be making sure that we build this, um, enterprise knowledge intelligence layer, uh, for companies to truly use AI agents, uh, in a trustworthy way. So, you know, our Stack Internal product that we launched a couple weeks ago at Microsoft Ignite, in fact, uh, is, you know, very, very excited about that on the on the enterprise side, uh, as well as, of course, on the public platform, as I've mentioned throughout, to help our community users, um, connect with each other, um, really learn as well as grow their careers.

And there are gonna be so many avenues and new entry points like our AI assist and our subjective content and chat and other things that people hopefully find very useful, uh, as things change around them very rapidly, and, you know, they can be part of this amazing community and and help each other out. So those are the two focuses, enterprise as well as our public community.

All right. Well, when the bubble pops next year, we're gonna have we're gonna have you come back, That's And we're gonna see you predicted it.

Thank you, Nilay, appreciate it.

I'd like to thank Prashant for joining me on Decoder and thank you for listening. I hope you enjoyed it. If you'd like to let us know what you thought about this episode or really anything else at all, drop us a line. You can email us at decoder at the verge dot com. We really do read all the emails. You can also hit me up on Threads or Blue Sky. You can also leave a comment on our fancy new YouTube.

You can watch full episodes at Decoder Pod, and we have a TikTok and an Instagram. They're also at Decoder Pod, they're a lot of fun. If you like Decoder, please share with your friends and subscribe wherever you get podcasts. Decoder is production of The Verge and part of the Vox Media Podcast Network. The show is produced by Kate Cox and Nick Stat. It's edited by Ursula Wright. Our editorial director is Kevin McShane. The Decoder Music is by Breakmaster cylinder.

We'll see you next time.

Life can be messy and complicated, especially during the holidays. One thing that isn't switching to the Simplicity Plan from Verizon. It's one plan, one price. Just $30 a line per month for you or the whole family. Make the simple choice this season. Switch to Verizon and shop their best holiday deals today. Why can't we get AI to production? Our competitors are all ready. How do we keep data secure? AWS AI cuts through the noise. Ready to use agents and the broadest set of AI tools.

Stop overthinking, start building. AWS AI is how. From the creators of Zootopia and Frozen comes Disney's Ha-xt. Welcome to Hex and enter the world of witches, where it was foretold that an ordinary girl would become an extraordinary witch and save the world of magic. Disney's Ha-xt arrives in theaters this Thanksgiving.