WordPress VIP is the content platform you know and love, built for enterprise.
Connect with Brian on LinkedIn and X. Learn more about his work at his website.
Shoutout to user Don on winning a Populist badge for their answer to How to ignore certain MySQL tables when importing a database?
TRANSCRIPT
[intro music]
Ryan Donovan: Hello and welcome to the Stack Overflow podcast, a place to talk all things software and technology. I'm your host, Ryan Donovan, and today we are talking about the necessary evolution of structured data, metadata, APIs, all that sort of thing, as it will have to change in the era of AI agents. My guest for that is Brian Alvey, who's the CTO at WordPress VIP. So welcome to the show, Brian.
Brian Alvey: Thanks. I'm excited to be here. I've kind of been a Stack Overflow fan for years, so it's going to be cool.
RD: We love to hear it. So before we get into today's topic, we'd like to get to know our guests a little bit. Can you tell us how you got into software and technology?
BA: I didn't plan on it. I didn't study for it. I think it was kind of waiting for me to find it. So I would have bet anything when I was a kid that I would be a comic book artist at this point in my life. And when I got to college, instead, I got an astrophysics degree because my dad said I'm not going to pay for an art degree. And so I went the other direction, which is like hardcore rocket science. I got back to New York City and I'm like, what do I do with this? And I worked in publishing for a bit with magazines. I was an art director and did a lot of support, a lot of technology stuff, installing networks, repairing computers. But then suddenly the web happened in ‘94, ‘95. And I was lucky enough to build two big websites in 1995, which was TV Guide and then a couple weeks later, Business Week. And so I've only worked with brands you've heard of, whether it was in my sort of solo career or now at work as VIP, where we work with NASA, the White House, Rolling Stone, New York Post, The Onion, Savannah Bananas, like just really cool brands. So I'm very lucky. But just solving technology problems. I don't know how comic books and physics gets you into that, but it turned into like the CMS career.
RD: What are the sort of new complications in the last couple of years for software has been AI and AI agents? It's kind of upended a lot of how we think about software development and software operations. Obviously, a big part of this is reliability and context. People are trying to figure out how to draw context into the agents. The pitch was that the way that the structured data, metadata, APIs are operating will have to change to reflect how AI agents work. Can you tell us a little bit about how you see that?
BA: There's seven different ways to talk about AI, right? There's a threat of AI stealing content from publishers to feed the engine, to be able to do the magic that it does, to make images, to make emails for you, to do all these things that it writes, all the creative engine stuff. But then there's also the business of building software, right? Completely upended that. Then there's just business in general. And it's completely upending that in a different way. So if you just take websites, I built a couple dozen CMSs and WordPress powers, you know, 40 plus percent of the web. So if you just take software that runs websites and you look at the behind the scenes team, right, the talent making the website, and you look at the customers and visitors side of things, just those two areas, massive transformations, massive disruption. On the building websites side, a long time ago, the only way to update a website was with code. So those sites in 1995, I worked on. There were no CMSs. You made code. You had to put things together by hand. Later on, CMSs and software. After a while, you're like, well, how does a friend of mine who doesn't know how to code change a website? You build a CMS. You give them a form. They can drag images in, type a title, hit save. CMSs, easy. But now, anybody can code with these engines, with Claude, with Codex, and all these things. And so now it's transforming how websites are made. And I think it's also, the next five years are going to be even crazier than the last five were. In terms of what survives, what software is needed for the web. So I look at the behind the scenes stuff, our business, the software we sell, the software we manage. And I think, all right, what survives this transformation? Well, I don't think you're going to need a CMS for a 20 page website or a 200 page website anymore, or if one, two or three people are working on it. But I think you will still need very well-built software if you're doing one of two things. One is commerce. You will always want great commerce software to run a store on, right? You run your business on that. It's important. The other is high-end media. So if I have 20 million articles and a 2,000-person newsroom, okay, I need something a little more complex than a vibe-coded GitHub-run markdown file-based thing. Okay, that's behind the scenes. The other side, which is, I don't even think was part of your question, but it's something I worry about a lot because we talk to our customers, is just how online businesses are being completely disrupted. So you think of the traditional sales funnel, right? You have the marketing funnel. If people come to your website, the top of funnel, middle of funnel, they get warm, bottom of funnel, they buy something. That was, I'd say, nibbled at by search and by social where maybe two or five or 10% of it was sort of disrupted, but then people still land in this funnel and they go through the journey. Well, the LLMs and answer engines are destroying that. Entire funnel is gone and somebody shows up already knowing what kind of car they want, what price they're willing to pay and where it's in stock. The bottom 5% of your funnel survives. So if you just think of business transformation, massive disruption. And if you think of the powering those businesses, the software side of things, like you said, the people that work on these things, the CMS world, also massive disruption and more disruption ahead.
RD: The new funnel, like you said, very disruptive, but it still needs to pull information from the business websites, right? I know that a lot of websites have complained of sort of being treated roughly by AI agents on the internet, by bots. Do you think that the way that they're serving that information will have to change for AI agents?
BA: A million percent. So it already has to have changed, right? Everybody who's not in the middle of changing is just so far behind. So what I think of is stepping back in like a first principles and just thinking about businesses, right? Forget the websites. What's your business? What's the business you're running? That's the first thing I ask somebody. Are you a marketer with a brand? Are you middle of funnel, right? Are you a media company, ad supported? Are you a commerce? Are you government, public sector? These are kind of the things that we work with at VIP, the enterprise websites. What's the business you're running? What do your conversions look like? Tell me what your business is. And I think that if you just look at that, the business itself doesn't change. So what you sell and who you sell it to doesn't change at all. I make sneakers. I sell them to people who have feet. I make cars. I sell them to people who drive, right? Like whatever those things are. Those things don't change. What changes is all the trappings of the whole machine of how that gets done. Now it's actually that person has an agent doing the shopping and I got to have an agent that's online in the middle of the night doing the negotiations with them. So really you go from 30 years of building websites for Google. Every stupid tag thing you did, just loading fast enough. What's this meta thing? What's it? Am I jamming enough things into the description? Right. Am I gaming their little game? Right. Am I trying to like win their little game? And you're literally just building sites for Google that people happen to visit, but still people visited them. The future is I'm building them straight up for agents and people don't visit at all. You talk about Google zero. What happens when you go to people zero? That's crazy. But that's how this is going to work. I mean, you're automating this sort of like buying and selling thing. But the business doesn't go away. It just completely changes how it runs.
RD: It almost makes me think that the future would be just one CMS to rule them all, right? Everybody just puts their articles in this one CMS somewhere and AI agents just draw from it, right?
BA: So I don't know that it would be the way you've said it, right? Which is there's one CMS to rule it all or anybody uses that software. But if it all just does end up in ChatGPT, the same way everything kind of ended up in Google and Google had a shadow copy of every website on earth. You could just go to Google and ask it anything. It has all the content, right? They didn't have to send you traffic. They just were trying not to be evil. But if you look at the ChatGPT thing, that sort of black hole, the LLMs, the big three, the big four, the big seven, however you count them, they could just be where all the information is. And that's really tough. And so you go back, you look at things like Stack Overflow, or a Reddit or anybody ad supported who doesn't have maybe subscribers or they don't sell sneakers. There's not a transaction happening on their site. There's an ad click and it's a few hops away to somebody else's transaction downstream. And you look at those and you go, yeah, how does that work? How do you get a direct relationship with your customers? How do you get customers out of businesses, communities of content?
RD: So in terms of, you know, sort of content and storage and serving to a website, what is a CMS?
BA: There's a couple parts to it. So there's sort of like a front end and a back end. One is, so a CMS is just some number of posts, some number of articles, pile of content. It's a database. And then there's a website. The part that makes it really a CMS is the part where there are specific bespoke tools for a team of people with special access to change the website without having to do code, right? Tools where you can go in and drag in videos and clip things and change things and change a headline and different things work a certain way. That's really the authoring side of things. Those 11,000 person newsrooms in London that we deal with, like the extreme high end of authoring on the web. That to me is like the best use of a CMS is managing all of that, the interactions between those people, the machine that runs, you know, the UK Sun or something like that, or some big, you know, Sunday Times. CMS is really about the staff, the editors, the team behind the scenes. But at the same time, that same software usually also runs the front end of your website. So it is the experience that people have. Do I see one story and then I keep scrolling and it just infinitely adds more things for me to see as I scroll down? Or do I have to click and go through 10 things to get there? Is it a game? Can I like things? Can I comment? It's all the audience features and it's all the editorial features combined. And that's CMS. And if you're talking about like, where does that all go? It's a funny thing because one of WordPress's strengths is the block editor. So when you're authoring a story, instead of it just being like a bunch of HTML soup, just a bunch of junk. Actually, it's very structured. You know, your headline is a certain block. WordPress has been very careful and cautious to make sure that they followed what were called web standards back in the day, back in the 90s and the 2000s. When you made something in the CMS and you output it, it's valid markup. That's machine readable. It's accessible to people who have good color contrast, like vision problems or hearing problems, whatever. So WordPress has always cared about accessibility and about web standards. And it's always been baked into its DNA to make good, clean, machine-readable code that passes validation. And it's funny that all the stuff they did for people who maybe, for screen readers, for the blind, or things like that, for all these years and made sure they always cared about people that were usually an afterthought in another product, that's actually paying off because agents– also kind of blind, you know, they can't hear, right? Like there's a way that agents work that the block editor, all the output is actually kind of primed and perfect when bots visit your site.
RD: The bots and the agents, they don't need all the fancy JavaScript on the front end.
BA: They don't care.
RD: I hear a lot of, you know, CMSs, databases, even some SaaS companies are going headless in the face of agents. How do you think of the WordPress viewpoint of like, is headless the thing you want to commit to at some point?
BA: The way you're asking it sounds like we are maybe not headless or we're a full stack thing, but like WordPress works great headless. It's had an API for, I don't know, a decade or more. So the way that bots see a WordPress website looks nothing like the way a person sees a WordPress website. So it's already ready for that. WordPress is a great backend. One of the biggest media companies that we work with, WordPress is their authoring and it's the CMS, it's the corpus of content, it's the 20 million records, the 200 million photos, it's all this stuff. But they have their own custom front ends or many custom front ends in many languages throughout, you know, the region where they operate. Same thing, one of our customers in London, same thing. Anybody can kind of put anything they want in front of WordPress. Now there's a caveat. The second you go and make your own front end for humans, the second you do that, you lose certain things. If you just use WordPress front and back, you get instant previewing. How you authored it is how it looks for people. If you go and make your own front end and go headless WordPress, you're then recreating previewing, which is kind of annoying. And you're also introducing security flaws because you have to pass draft content, hidden private content, not yet published content, embargoed content to a front end somewhere. That's always going to bite you. So while I'd say headless gives you a lot of headaches, headless is very easy to do with WordPress.
RD: I remember playing around with WordPress sites when the blocks change came about and there was a lot of outrage over it.
BA: Outrage is a good word. Yes. There still is. There still is.
RD: Still is. Well, because it was a different paradigm. And I wonder. That seems like that's a sort of comparable change that affected the backend storage a lot.
BA: Yeah, it did and it didn't because they really kind of jammed it into the old structure. So they actually pack blocks with comments around them back into the old post format that's been around for 23 years. So sort of nothing changed. You know, it's these hints around where blocks are. They didn't fully commit to blocks in such a way that it destroyed everything. In my opinion, as a person who was not born in open source, they're too backwards compatible. They worry too much about the community. They worry too much about making sure 60,000 plugins of various quality continue to work with every new release. That's wonderful that WordPress does that. I think it's above and beyond. I think it may be too much. Very generous.
RD: People build their business on WordPress, right?
BA: No, no, 100%. So talk about the block editor. The block editor is a funny thing because they introduced this and really it was about the moment of authoring, just making sure things are structured. We have customers, massive media sites that you know, big brands, who are like not on the block editor yet. And it's been out, I don't know, five, six, seven years, like a long time. And it's kind of crazy. And just the way that you used to store HTML was kind of like a junk drawer of just throw all the stuff in there. And the block editor is kind of like the drawer in the kitchen where they organize the spoons and forks and knives and they're in little slots, right? It just kind of puts a little framework around it, little borders around things, right? So it doesn't really change it that much. And actually the block editor to me is probably the most powerful and beautiful and slick feature of WordPress. But I don't know that it matters going forward. If I'm just dictating to Claude while I'm driving, does anybody care? I think the block parts survive, but I think the interface for dragging, you know, like a Google Docs style interface for, you know, real time collaborating on these blocks. I think humans typing things in a browser, if that goes away, okay, the block storage is good, but do you really need block authoring, right? I think that being a strength of WordPress is interesting going forward because I don't know that people typing into a browser on their laptop is going to be how you even do the New York Post or Rolling Stone or one of these big media companies.
RD: I think that it's just going to be a series of MCP servers just passing text around.
BA: I really do. I think it's personal agents. I think anybody who's going to survive the high-end newsrooms, 1,000 to 11,000 person newsrooms that we work with, I think that they have to go from people are doing a thing and bots are helping every now and then, kind of like Grammarly, you know, helps you, but it doesn't do the work. It does 2% of the work. It just makes sure you don't look stupid. You don't type something wrong. You don't typos, right? You don't say the wrong thing. But I think it has to flip. I think it has to be the machine does most of the work. The machine understands your brand guidelines, your tone, your voice, or whatever. It understands the intent of this piece of content. So writing up a recap of a New York Knicks game or a World Cup game is one thing. And there's a prompt for that. You know, you would call it a brief, you know, I'm assigning Ryan the story. Here's the brief. Here's your marching orders. Well, that's a prompt going forward. And I think you need to give that to machines that then do the work, they assemble, they pull the stuff together, and they run it by a human. And they have to be like your super coach. They'd be like, okay, Ryan, okay, I got a couple headline ideas, but I know you're really good at this. Come up with something really funny for the front page. The New York Post has the best pun headlines and stuff like that. So there's still some creative human elements, but think of how you use AI today, to be better at your job, to write things faster, to be a thought partner when you're writing something sensitive for somebody and you want to make sure you didn't leave anything out and you just want to get a second opinion on it without having to wait for a person. That’s really cool, but now flip it and make that a super coach, a super editor who's telling you, okay, what about this? What about this? And you just fill in the pieces it needs. I think you have to flip to that to really survive the work model going forward for building content.
RD: For little sites, because of the ease of sort of vibe coding or spinning up a sort of website, they may not need a CMS anymore, but the ones who do have those massive organizations. What's the benefit that they get having that sort of content data store?
BA: I think there's just different rules, guidelines, just guardrails and things like that. So we run government sites like White House and NASA, right? So we're FedRAMP authorized. We went through all this hell to become authorized, to be able to show that we're secure and we scale and all this stuff, right? We're like government approved. So they have a lot of rules around what they do and can't do. So if you think of certain things like, you know, a government contractor has access to one of those sites I mentioned. And then that government contractor is like no longer working with them. You want to make sure that they're out of the system instantly. Or even if it's Rolling Stone or, you know, the whatever, The Onion, right? Like you want to make sure they're out instantly. So there's a lot of those things that just don't matter when you're vibe coding a 200 page website for your software brand or your whatever it is, your technology company.
RD: There's less policy to encode into the software, right?
BA: A hundred percent. And so really somebody sent me something years ago about like, engineers working together and how like, managing a two or three person team is okay because there's only so many like connections between the nodes. But once you have 30 people, it's like a billion. And then once you have a hundred people, it's like a trillion. It's just, you know, the numbers are stupid. And so that same kind of complexity comes in when you have these multi-layered newsrooms across different time zones, working different sections, and you have thousands of pieces of content to crank out every day. You put powersalesforce.com or Jefferies and like banks and things like that. They don't have the volume of content of one of these newsrooms, right? So they may put out 30 things a day, 30 things a week. The thing that I love about what we do and who we work with is it's extreme WordPress. Somebody has the largest database that anybody's ever had. Somebody has a terabyte database for their WordPress website. And any place, if you're hosting that yourself or hosting that somewhere else, like they're like, they throw their hands up and they're like, yeah, we can't help you. The backups, the hourly backups take four hours. We can't help you. Like that's insane, right? The math of it, it just doesn't work. So, or they have the most DDoS attacks or they just have the biggest team or the compliance around, you know, audits and who signed in and who edited the headline and who gets yelled at for something when we get sued, right? So there's just in many different ways, whether it's traffic, whether it's attacks, like I said, there's just each customer, you know, has a different extreme WordPress challenge. And so we learn a lot from that. It's a lot of fun. And you bring it down to a team of one, two, or three. But think about that. If you made a site yourself, if you made software for yourself, and you ran a website, a couple hundred pages, a couple thousand pages, you're good. The second you have somebody else working with you, I'm like, ah, I have to explain this thing to them. How do I stop them from screwing it up? Now times that by a thousand. And you're like, all right, I give up. I need software that's been doing this for decades. I need somebody who talks to customers bigger than me on a daily basis so that they're solving the problems that I'm going to run into two or three or four years from now.
RD: What's one that tested the sort of limits of the software?
BA: I don't know that there are a lot of software limits or even hardware limits per se. I think of business cases. There's one customer, they basically decided to run an auction site in WordPress. So like, I'm going to build eBay on top of WordPress. And they're like, all right, interesting, weird, right? Whatever. You can do anything you want. It's PHP and database tables. Anything you've ever seen, you could build in WordPress. Maybe you shouldn't. But you could. So they did this thing and it just took off like it was really, really, really big. It became huge. And what they said was, OK, a blog post is an auction and the comments would be the bids, the bids on the auction. But they would have like a billion comments and it was like real time bidding. And like the software wasn't built for that. So they kind of came to us like, hey, this isn't scaling and we have a lot of trouble. Can you help us? And I cannot say we were like, flawless, because you're building eBay on top of WordPress, right? Did a lot of things for them and with them. We like, optimize things. Okay, kick the bids out of the comments table because it was never built for that. You know, here's what you need to do. So we've worked with them on it. But I think it's both a, I mean, it's a success story, but it's also just like it was one of the hardest things. It's just not easy. You know, it's very hard to do. So that's one of the customers I think was just, wow, why would you do that on WordPress? But they did. And then it just got so big. Like they can't just go switch, right? It's hard to switch or build a new system. And they're at a big company and a big brand. So we just keep making sure it works.
RD: It's interesting, like that one seems like you're abstracting like the idea of what WordPress sort of originally was to a whole different situation. And it sort of goes back to the agent conversation where the agents are now sort of abstracting WordPress back to like just a sort of text database almost.
BA: Yeah, but one great customer who you just throw .md on the end of any URL and you get the markdown version. That's a peek at the future. I think where you need to go past that. is not just serving up the blocks in markdown format, which is basically what they do. If there's 30 blocks in a post or three blocks in a page, serve them up as markdown. That's cool. But I think putting in the hidden secret prompt saying, here's a car configurator. Here's how to buy a minivan. This is the Toyota page for picking between this, this and this. I think putting in things that help because you have to assume that that experience is no longer going to live on your website or if an agent's taking it, it's running it in a chat somewhere. So now you're showing up in some chat somewhere asking something and it's got to tell you which minivan is the right one for you, or which number of seats, or which– is it all wheel drive or whatever. So I think having websites that are able to not just give you machine readable stuff but actually give you an experience, give you something that makes you look good so people choose one website over another to get their news. Like I could read about the World Cup on a thousand websites but I'm choosing to read about on this one, right? Why is that? Why do I love that brand? Why do I love Apple? Why do people love Harley Davidson? Why do I love a certain brand of ice cream? Why do you love this thing? How do you take that same brand love, that thing where you're going to rave about it? Oh my God, I got an air fryer. Let me tell you all about it. It changed my life. People talk about these things and it's crazy. How do you get something, a product or a website to be so good that bots will tell other bots that agents are going to go rave about this thing over on Maltbook or something like that, right? So I think about what are the signals to an agent that say, if you're going to use a website or a website software, this is the one you should use because the other ones are inferior for these reasons. They're human signals. I love the logo. Apple products are shiny. It looks rich. It makes people like me. Whatever that is, how do you do it for bots? So I don't know. I think that's one of our biggest software challenges. If we go to human zero. How do I make sure that bots run towards our websites and not somebody else's?
RD: That's the new sort of SEO is trying to do an inception on the bots there, right? Do you think the effectiveness of that is going to be a sort of data structure problem? Either, you know, let's say equivalent to a sort of text problem.
BA: So I don't– I actually don't think it's going to be about the content or the presentation of the content or that structure. So I think the APIs are there. I think the ability to change– change blocks into Markdown or whatever it is. I don't know if that's better than HTML or not better than HTML, but I think that's a good first step and it'll make people at these companies feel like they're doing their job. I'm ready for agents. But I don't actually think that's what agents want. I think agents actually need a completely different experience. So if you think about shopping on, you know, Toyota.com and I'm looking at kinds of different cars and doing a competitor, there's a whole visual thing that's happening for a person who's looking at cars. You're doing aspirational things. You're helping me show three of them side by side. You're helping me do a 360 view of the car, you're doing overhead to say how many people fit, whatever that is. That whole UI for humans is one thing. It's a relic, it's, it's for humans looking at websites. Let's say that goes away. But if you think about the agent side of things, now you think, what are you building for them? If I go to, you know, toyota.com/sienna/whatever whatever, that same thing is, I shouldn't have to click around a bunch of web pages. I shouldn't have to load a bunch of things. I should just get assembled a bunch of structured data in some big format that explains to me, here's what people need. They're looking for this. Here's what states need all-wheel drive and what states don't need all-wheel drive because there's no snow. Here's the different versions of the car you can have. And basically explaining to the agent how to then surface that in their ChatGPT, their Claude, whatever their interface is where people are in there, how to look smart. So if you can make an agent and if you can make a chat bot specifically look smart, your site is going to win. Here's what would make a chat agent look smart. If I go into a chat bot, I go into ChatGPT or one of these things, and I go, hey, I'm looking for a minivan. I'm looking for a car for my family. And they go, oh, you probably want a minivan because I know from previous two years of you chatting with me, I know how many kids you have and how many dogs you have and all this stuff. Here's what you need to think about when you're buying a minivan. And if you can give from your website the right information to that chat bot so the chat bot can say, hey, I figured this out. I think if you need seven or eight passengers, that would be this car. Here are three quick next questions you might ask, which is like, show me a dealer near me. When can I do a test drive? Show me a 360 view. And if you can prime the chatbot to know the three questions that a car buying shopper will ask, the chatbot may or may not know that natively, what questions a car buyer might ask. It's hoping that the website for the car buying will actually surface that. And if you can be the one car buying website that does that, you'll be the one that it offers. It's only going to talk about Toyota. It will never talk about Chrysler and all the other minivans. So that to me is what I mean by like, look good for bots, make bots look good, and you win.
RD: So if you could just give them a response they can copy and paste?
BA: You have to feed them that. No, no. So think of the Steve Jobs thing when he introduced the iPhone. You know, journalists, especially bloggers, are like, don't tell me what to write about. If you tell me to write about that, I'm not going to do it, right? They're like, you can't push me, no. But Steve Jobs said, hey, I'm introducing the iPod, right? And it's a thousand songs in your pocket. And like every headline in every newspaper for the next two days is like a thousand songs in your pocket. He basically told them what to say, but he made them look good. That's a cool story. That's a great headline. So he's helping me do my job. So if you can do that where you're helping the bot do its job, when somebody is asking it a question, you help it then take the person down the path to have a very smooth experience. They're going to give a thumbs up to the bot. The bot's going to go, oh, I'm totally going to keep recommending Toyota over Chrysler.
RD: So it sort of also makes me think of, you know, people have been talking about the semantic web for decades, right? And it's never really caught on. Do you think this will lead to ever more structured text? Or do you think that the LLMs are good enough to sort of parse it out?
BA: So I think both. I think at the beginning, giving it the structured text while they're getting smarter or whatever. But like once you reach, you know, Fable and beyond, right? Fable 5, Fable 6, Fable 16. It doesn't matter. I mean, it is funny how many things we think around prompt engineering or things that we have to do or, oh, bots are dumb, so we have to do this thing for them. Okay, the bots are dumb today. They're actually not dumb. They're magical. They're insane. They're life-changing. They're smarter than we all are. But the bots that we're dealing with, the agents that we use today are the dumbest agents we're ever going to use again for the rest of our lives. Like tomorrows, next weeks, next years. You just have to assume. They're all going to get better and they're going to all answer all the questions. And I think that thing about like serving them Markdown because it's digestible for agents. Markdown is just a human shortcut notation for HTML. Like that's all it is. So there's no, you're not saving anything by giving them Markdown necessarily. Somebody actually found it in terms of saving things and saving tokens– somebody found a really cool thing, which is I can give a screenshot of like a book's worth of text to an agent and I'm sending it one image and I'm only paying for like three tokens or whatever. And it goes in like OCR is the whole thing and figures it all out instead of sending it 10,000 words. So there are ways to cheat and prime that. But I think the whole argument of like Markdown versus HTML versus whatever, they're going to figure it out. Whatever your hacks you're having to do to feed something to the bot because they're not smart enough, that's this week's problem. That's not next week's problem.
RD: Before we hit the road, I know you all put out a future of the web report. We are definitely concerned about the future of the web, Stack Overflow. What did you find?
BA: It's good. We surveyed 2,000 people. I think it was like 12,000 enterprise buyers and like 800 people who use AI or vice versa. I forget which one. But basically, we found out that people don't trust the answers yet. So they do click through. 86% of them click through to the websites to verify things. And I've seen this in my own life. I'll give you three examples. So one is I asked something about people in a movie who had the same birthday as me. And it said, oh, I read IMDb and here's the answer, but actually mixed up two people. It took like a main character and then the minor character had my birthday and like put them together and gave me a lie. But it's like not ready yet. Like that's a bot saying, I just read IMDb. So you don't have to, here's the answer. And it was wrong. I also look at chatbots and I will tell you that most of the chatbots that I deal with can't do the thing I need. So on like an airline chatbot, I'm trying to do something like,
just give me an agent because I have to do a thing that's more complex. I've only seen one chatbot where I went to do like, a product return, or they're supposed to send me two of something. Only one came in the packaging and a chatbot like, helped me start to finish, told me what to do. Like I've only seen one chatbot actually nail a complex problem. So we're not there yet. And so what we found is people click through, they don't trust, they want to go read about these things. And I think it's like in the, remember when you first got Gmail and everything went into spam and you still went into spam and you looked. Because you're like, wait, is it right? But it took you like a couple weeks, maybe a few months. And then for the rest of your life, you never look in the spam folder again. So we're in that early day where people don't trust. They go through, they verify, they look at the thing, they click the links, they read. Okay. And at some point it's going to flip where you're like, I just know that it reads better than me and it no longer gives me wrong answers. They fixed all the problems, but we're not there yet. So that's the biggest thing from the report is that people don't trust the answers they're getting yet, but they will. The reason we did the future of web report was because we have these massive customers, these global brands that are worried about the future of the web. We're worried about the future of the web. So we did the future of the web report and tell you the best thing we have. What we do isn't really like unique software or the best hardware or scaling or whatever. It's just talking to these people. So we have access to the biggest customer problems. And therefore, like I said before, you know, we know things if you're a smaller site you're going to hit these eventually we're hearing about them because we have like a customer advisory board where we sit in the room with like 12 people from media who are running into the biggest problems or 12 people from– so if they come to us and they say, you know, MCP doesn't matter. Great, we won't build MCP. But if they come to us, like, we'll know within three months I need MCP, I need it to be secure. So one of the– we just launched– two or three days ago we launched a secure MCP for our whole platform for Wordpress and all of our analytics. So you know it's, um I think it's the having a seat at the table on the cutting edge of the biggest problems is probably the most important thing. And that's why the future of the web matters, because everybody's scared.
[music]
RD: All right, folks, it's that time of the show where we shout out somebody who came on to Stack Overflow, dropped some knowledge, shared some curiosity, and earned themselves a badge. Today, we're shouting out the winner of a populist badge, somebody who dropped an answer that was so good, it outscored the accepted answer by 10 points. And today, the winner, we're shouting out Don, who answered, ignore certain MySQL tables when importing a database? Curious about that? We'll have the answer for you in the show notes. I'm Ryan Donovan. I edit the blog, host the podcast here at Stack Overflow. If you have questions, concerns, topics to cover, you want to reach out and say hi, you can email me at podcast@stackoverflow.com. And if you want to connect with me directly, you can find me on LinkedIn.
BA: So where am I? WPVIP is the WordPress VIP. WPVIP.com is our site. BrianAlvey.com for everything else. Find me on LinkedIn. I'm on X. I'm all over the place.
RD: All right. Well, thank you for listening, everyone. And we'll talk to you next time.
[outro music]
