336: What Comes After Code? JAMstack, AI Agents, and the Open Web with Matt Biilmann, Netlify CEO
37:15 Share

But I do think that we're going to have a lot more different user agents interacting with those URLs and a lot of those will be at Yentik. And I think we have a lot of work to do to figure out how we build the user interfaces and the agent experience on the web for the user agents that are acting on our behalf instead of directly through us.

Good morning, good afternoon, or good evening, depending on where you're listening. Welcome to AI and the Future of Work. I'm your host, Dan Turchin, CEO of PeopleRain, the AI platform for IT and HR employee service.

Our community is growing thanks to you, our loyal listener. If you haven't yet, sign up for our weekly newsletter. We provide some extra AI fun facts and some clips from the show that don't always make it into the main episode. We will share a link in the show notes. If you like what we do, please tell a friend. Give us a like and a rating on Apple Podcasts, Spotify, or wherever you listen. If you leave a comment, I just may share it in an upcoming episode.

Like this one from Warren in Hartford, Connecticut, who is in software sales and listens while enjoying his daily ice plunge. Good for you, Warren. That's you and not me. Warren's favorite episode is the great one with Stephen Messer, founder of Collective Eye. I love that one. From back in season two about his journey from attorney to AI entrepreneur.

We will share that link in the show notes as well. We learn from AI thought leaders weekly. And of course, the added bonus, you get one AI fun fact. Today's fun fact. David Jenkins writes in Design Rush about the future of AI in web development. In it, he says the market for AI in web development will grow at a compound annual growth rate or CAGR of 25.2% between 2023 and 2033.

He says AI tools including GitHub Copilot, Tab9, and even Figma are increasingly used to automate coding tasks as well as generate content, improve UI and UX, and test and debug sites. Three significant advancements he sees are NLP that will make it easier to navigate the web, improving accessibility for disabled users, and also content that becomes instantly available across all cultures and languages.

My commentary is kind of a tired theme. Everything in this article is obvious, at least to this very informed audience. What's more intriguing is how the web itself may become obsolete. Agents and spatial computing may fundamentally change how humans and even non-humans interact with the web. Keyboards and browsers may become artifacts of the past, relics from a time when we had to adapt our behavior to how machines work.

The future is just the opposite. Machines should be smart enough to adapt themselves to how we work. This is a topic that's very germane to today's conversation. Of course, we will link to the full article in the show notes. Now shifting to this week's great conversation.

Matt Bilmon isn't just the outspoken CEO and co-founder of Netlify. He's one of the most influential leaders in the open source web dev community. To date, he and the Netlify team have raised more than $200 million from an incredible list of investors, including Bessemer, A16Z, Bond, EQ2, Kleiner Perkins, Mango Capital, and Menlo Ventures. Full disclosure, I'm an investor in Netlify through Mango Capital, where I'm an LP.

Billman's journey began before the rise of cloud computing and AI in 2014. He and Christian Bach, his co-founder, started Netlify, driven by a vision to simplify web deployment and unlock the potential of the JAMstack architecture. Matt received his bachelor's in, obviously, musicology from the University of Copenhagen, which, of course,

is the birthplace of Little Mermaid author Hans Christian Andersen. Matt was a serial entrepreneur before founding Netlify. And without further ado, it is my pleasure to welcome Matt to AI and the Future of Work. Matt, let's start off by having you share a little bit more about that illustrious background and how you got into space. Thanks, Dan.

Yeah, I mean, somehow I've been building developer tools for the web for decades now. I'm from Denmark originally, but I spent seven years in Madrid in Spain before I made my way to San Francisco in the Bay Area. And there I worked as CTO for a company that had a lot of different business units. We ran...

the second largest social media network in Spain, the largest hostel booking site in Spain. But our core business was building websites for small to medium businesses at a very large scale. So at the peak, we would build around 100 websites a week.

And my teams would build the whole platform that the designers would use to do the design with, the clients would use for content management, and that powered every single website all the way from the first brief into production hosting. And then based on that experience, I co-founded a company in Madrid together with the founders of that company.

Sort of saying, let's take this kind of platform and build a multi-tenant cloud-hosted CMS that other professionals can use to get the same efficiencies when they're building for their clients. I came to the Bay Area working on that. And it was while I was working on that that I sort of started...

building the belief that there was a big need to change the core architecture of the web. Well, almost all websites, web, like e-commerce, web applications at the time, they were typically built as monolithic applications where you would have like the data layer, the business logic layer, and the UI layer all built together as one monolith. And I started sort of seeing like the very first headless,

content API start to emerge and believe that that was like the way to go, but also believe that like the flip side of that was like this decoupling of the web experience layer into its own thing that could use web native tooling and be built independently. And

I ended up coining this term JAMstack to describe that new architecture of decoupling the front end from the back end and using a bunch of different headless APIs as the back end and sort of really building with a headless architecture. And Netlify was sort of really built with that architecture of mind, right? In a sense of like, if we thought...

That that way of building would be the right way to build for the web, to make websites faster, more performing, more secure, better user experience, better for developers to build and maintain and so on. How could we build a platform that had such a great developer experience that developers

that it would help developers adopt this architecture and thus help build a better web, right? And that really became the starting point for Netlify. And we launched in March 2015, almost 10 years ago now. Just me and a co-founder, Chris, bootstrapping the first year before we then raised the first round of capital and then several rounds after that.

Today, we've onboarded more than 6 million developers onto the platform and we are reaching more than a billion unique visitors to the site stores and apps running on our globally distributed edge network and serving customers of any kind from hobbyist developers to Unilever or Riot Games or North Face and Vance and big companies like that.

So for those uninitiated, you can roll back the tape a few seconds. And in fact, you did hear Matt say he is the one who coined the term JAMstack, which is now a bit of a household word in the web dev community, acronym for JavaScript APIs and markup.

So Matt, you had that insight a long time ago, 11 years ago, or maybe longer actually, and popularized a whole new way of thinking about web development. The company has changed so much in the last 11 years. What was the biggest inflection point for Netlify that led to the difference between when you launched and where you are today?

It's hard to say what the biggest inflection point were. There were many different points around the journey. I will say that one point was sort of like the point where I publicly coined the term Jamstack. I gave a talk at a conference run by Smashing Magazine in San Francisco right at the start of 2016.

where I sort of first really like publicly used the term and introduced that new stack. And that in itself led to Smashing Magazine, which at the time was probably like the most popular online resource for front-end developers to come to me and be like, hey, how do we adopt this Jamstack technology? And we then helped them sort of

think through all the paths they were running, like their main magazine on WordPress as a monolith. And then they had another Shopify monolith for selling books. And then they had another monolith for their events. And then they had a job board built in Ruby and Rails and so on, right? Like, and they were thinking about a big redesign and like,

didn't know how they would execute that across all those different template languages and so on, right? And we really went in and helped, okay, this kind of complex project that both includes commerce and events and membership with logins and magazine and so on, how would we make it really straightforward to build that with this new architecture? And I think when that went live, it was sort of like a real proof point of like, hey,

we can do this, right? And this is faster and better experience and a better way of working and sort of was a big inflection point in the journey, led to our Series A fundraise and started sort of like, I would say, the biggest period of growth. And then another inflection point has just sort of been coming to the point where

In some ways, GAMSTACK is not something that people talk as much about anymore because now it's just kind of how we built for the modern web. Now we're just kind of in a stage where the theory played out and this architecture kind of won. And now the natural way of building for the web is with front-end first web frameworks in web-native languages and on platforms like ours. So that's kind of like...

a new stage of the company. And now I think we're then in front of like a whole new inflection point where I think we are about to witness very drastic change in how the web is going to be built once again because of AI, right? And that's kind of like the next big chapter in our journey, I think. So you have this revelation.

maybe analogous to the revelation about the JAMstack, that in the future, the web may be as much consumed by agents as it is by humans. Talk us through that new approach. And I know you coined a new term as well, but what's the future of the web? Yeah, I mean, I wrote an article called Introducing AX: Why Agent Experience Matters.

That's really reflecting on one of the things that became a key part of our AI strategy from, I guess, the start of 2023 was how do we make Netlify the best platform for AIs to build on? And in the start of 2023, that still seemed very sci-fi-like in a way. And now it's starting to seem very real. And I think...

We were sort of early in then thinking, okay, what do we actually need to do to make our platform better for AIs? And we started seeing coding agents emerge. We started building like a little GPT to...

deployed to Netlify from within the chat TPT ecosystem and it got integrated into some of the assistants that emerged there, like the Grimoire coding assistant and so on. And we just started organically seeing suddenly, like suddenly we looked at it and there was like a thousand new sites created from chat TPT every day. And it's like, oh, wow, something interesting is starting to happen there. And

As part of that, we built some specific workflows for AI agents to be able to initiate the deployment of a site before a user had even created an account with us or anything like that, and then hand over to the user after.

Then Devin from Cognition Lab made a big splash with their initial video about their vision of a coding agent. And it was again integrated with Netlify from the beginning and deployed to us. And at the end of last year, we saw Bolt.new and Lovable launch.

Bolt.new with a full Netlify integration and Lovable with an integration through GitHub that sort of really started driving large volume and that I think really represented, for me, a changing point where it got really obvious that now we're starting to get to the point where most new projects are soon going to originate with an agent that write the initial template, the initial setup and so on. Right?

And so I started thinking about what do we need to holistically do to cater to that and realize that we needed to build a discipline of AX or agent experience. Just like I mentioned, to make the Jamstack architecture the winning architecture, we had to focus on developer experience or DX and really focus on that holistic vision of how do we remove all the friction

from a developer that is writing some code and want to get it on a URL in front of a user. All the friction and iterating on that and getting deploy previews and getting it and operating it in production and so on. And the first 10 years of us was really the obsession of like, what's the simplest form there?

Now we are starting to obsess in a similar way around AX, right? Like what's this, if you're a developer and you have an agent writing most of the code, what's the path of least friction for the agent to understand our core primitives and how they work, to deploy, to Netlify, to get it in front of a URL, to manage that. And then on the other side of that, since we are a platform to like with a vision of building a better web, we have to start thinking about what about all of our customers,

that are building web applications or websites or e-commerce, how do they need to build their projects so agents can interact with them? And so they can have a great agent experience as well. And what is the path to make the web itself have a great agent experience to keep the web relevant in a world of AI agents and so on, right? So that's really become the next big thing

chapter for Netlify. And I think the next big fundamental change in both how the web is going to be built in the future

and how we're going to interact with the web, right? Like you mentioned it in your intro as well, right? Like the sense that maybe a lot of our interactions with the web will happen in new forms that are driven by agents versus like us manually sitting and clicking around in a browser. And how do we, like, does that mean that everything is going to be completely disintermediated and we're never going to like actually build something that a user interact with?

Or are there new ways to make the web really relevant to interact with? And what does that look like? On this show, we always talk about the human side of AI. And I think you and I both date back to a time when

As web developers, our IDE was Notepad. And we were intimately familiar with every line of code, every tag, right? I mean, we had to script our buttons and our forms. And so in a way, like modern web development, I mean, it's a crude analogy, but I feel like we used to be like Amish crafting furniture. You know, hand, right? I mean, we took a lot of...

pride in our product. And now maybe it's like Ikea building furniture, right? You don't have the same visceral connection to the product. So in the world where agents are building sites and

Maybe even agents are using sites. What does that do to the connection that the developer has, you know, and the pride that you take in the output of your code and people enjoying it? Just, you know, it seems like that's as fundamental a shift as anything we've seen. On the one hand, it's a fundamental shift. On the other hand, it's kind of like a very logical progression of like what's happened since, you know,

since we invented computer like human user interfaces and and like started out with like a punch cards as the first interface right like and then started building command line interfaces and then like build visual user interfaces came up with browsers and the web and like a higher level of abstraction right like and then in this jam stack period like if i think back to like

2014, 2015, when we launched Netlify, a developer building for the web was really mostly a backend developer, right? Like writing a little below in PHP or Ruby or Java. And then there was some element of frontend and HTML and so on, but it was mainly generated by the backend, right? Like

Suddenly we got this new layer of React and Angular and Ember and these native component toolkits and so on. And we had this period around the time when we launched and grew in 2016, 17 and so on, where suddenly all these bootcamps started spitting out React developers. And there was sort of this...

often this sentiment that these are not real developers, these are just like they only know React, they don't fully understand the system and so on. But I think it became pretty clear pretty fast that these developers were adding a lot of value by building at a new layer of abstraction without having to worry too much about those layers below, being able to just take different serverless API and services and

companies own microservices and APIs and then build great experiences on top of them. And maybe in that 10-year period, we maybe added like, I don't know, like 14 million new web developers to the world that came in addition to all the existing sort of backend developers that were there before, but that now started creating a lot of value, right? And I think

As developers start adopting agentic tooling that's going to be writing a lot of the code, we're probably going to see, in my mind, probably over the next 10 years, the next 100 million developers into the field that just didn't exist before and that operates on a level that's like a level of abstraction up, right? Yeah.

But it's still going to require a specific skill set and there's still going to be tons of knowledge you're going to need to have to use these adjunct tools to build great experiences with. And it's still going to be a rapidly evolving field where you constantly have to keep up, right? Like we just always have this layering up of new levels of abstractions you start working on. And I think this will just be another new level of abstraction that will make

their ability to build, like in some ways it will lower the barrier to building stuff, but it'll also just drive up the demand for building even more stuff.

I feel like the web never wanted to be confined within a little browser, like it's very artificial. And so through a combination of constraints of the technology and just constraints on our ability to be creative and innovate outside of the browser, we've kind of limited the web to what it is today. Those are just kind of my thoughts. You're the expert. Does that resonate with you, that the fundamental nature of the web may be

becoming obsolete or will change? It'll change. I think some of the things I believe will stick around is that are really fundamental to the web is probably like there'll be some like the web always had the concept of user agents, right? Like

different browsers that could read the information in different ways and present it in different ways, right? And it always had the concept of URLs as like the universal resource locators, right? Like that's sort of like the most brilliant thing, right? Like that

that you can just share a link to anything and send it somewhere and then we can see the same thing, right? But we might see it in different ways depending on the user agent we go there with, right? Those characteristics, I think, are really important, right? Like one, domain names and URLs, right? Like I can buy a domain and I can decide whatever the hell happens under that domain, what technology I build it with, what I do there, right? Like I can share URLs and instantly...

anyone in the whole world that I share that URL with can access whatever I put there and they can choose how they access it, right? Like they can choose which user agent they use and how they perceive that from screen readers to browsers to future AI powered devices or agents that actually go and interact with that URL on their behalf, right? Like those, I think are really sort of like the unique pillars of the web and

that will stick around. The other thing is that web has also become like an application platform with some well-defined programming languages. That means that you can distribute full applications on those URLs instantly. And that's another thing I still think will be useful, right? Like there's some people that seems to think that like if we have agents, they will automate all the workflows so we won't have apps

And I think that's slightly like, I don't know, like even if you have a ton of humans at your disposal, you don't just give them all phones or email to call each other or write each other. You also build apps that...

that are collaborative and show what the hell is going on and keeps trace of it and gives an interface to it. And I don't think that will change just because a bunch of the participants are going to be agents instead of humans, right? So I still think we will be building UIs. I still think that as humans, we crave that.

shared experiences, right? So I still want the ability to not just sit with my personal agent and have everything happen in my universe. I also want the ability to send you a URL and show like, hey, this was the product I was looking at. What do you think? Should we buy one of these? Or this was the travel destination I was thinking about. What do you think? Does it look good? All of those things

that happens between humans and that's carried by URLs and the idea that we have some shared mechanisms of how to perceive those URLs is still going to be important. But I do think that we're going to have a lot more different user agents interacting with those URLs and a lot of those will be agentic. And I think we have a lot of work to do to figure out

how we build the user interfaces and the agent experience on the web for the user agents that are acting on our behalf instead of directly through us. So you are one of the highest profile advocates in the open source community. So I got to get your perspective on this one. So you just shared a bunch of things that are kind of platitudes at this point in time about the brilliance of the

foundational building blocks of the web, URLs and protocols and all these things we take for granted, this phenomenally complicated system that makes the web work. So we kind of get comfortable with the idea that that piece of the internet is not owned by anyone. And that's a good thing. Enter the agentic world where there's just this

furious commotion of private industry and open communities. There's not yet resolved answers about who owns the agent-first web. What's your perspective? Is it just carrying forward the legacy of the open source community and how the web was built, or are we competing with some new forces here?

I think every time, like for context, when I started Netlify and we raised our first round of funding, our first round of funding was by far the most difficult to raise. And in part because most of the VCs would ask us, like, why are you building a web platform? Right? Like, isn't that kind of like it's all mobile apps now and the social media?

Right. Like the web's kind of like, it kind of didn't really last, right? Like, and it turned out like that if we gave people the right tools to build really cool stuff on the web, it could still become really relevant, right? Like, and a lot of companies and developers and people played a role in this. We played some role in that, right? Like, but it was important to keep the web a really relevant and compelling platform for people to use. And I think every time

there's a platform change, right? Like an AI in this sense, an agent in this sense is a platform change, right? Like away from the way, like just as big a platform change as mobile devices were, right? Like that sort of really gave an opportunity for rethink user interfaces and how we build and so on. Agents is an equally big platform change, right?

And every time there's a platform change like that, there's going to be a battle between closed platforms and open platforms every time, right? Like, because there's going to be a set of interests always on both sides, right? Like the interest on closed platform, pretty obvious, right? Like if you own, if you get to own the platform, it's very lucrative and very great, right? Like, and the interest on open platform, like,

Also, naturally, in the sense that a lot of us will try to figure out how we avoid having one overlord that in the end owns everything we do. But it requires every time people to invest in that open platform and there will always be certain disadvantages

from the speed you can move at and the need to come up with formats and protocols and tool sets that works between often competing companies and industries where closed vendors can just go build a closed thing end-to-end and take fast decisions. So it's always going to be sort of this dual business.

process where often you will see some initial shifts towards closed off walled gardens and then reactions on figuring out how do we make sure we keep the open alternatives competitive and compelling. And that I think is going to be just the kind of motion we are going into where it's going to take real work to make sure that we end up

with another version, the next iteration of a truly open, federated, non-controlled web that we can all do what we want with. And it's going to be really important. It's not something that's a given or something. It's something we have to go work for. So in that future version of the web, let's maybe call it, to use your term, the AX-first version.

Presumably, the infrastructure is going to be a lot different. And also, there'll be a new set of personas or roles. It won't look like today's set of web developers, just like today's web developers don't look anything like the web developers in 2000 do. Exactly.

Put on your prognosticator hat. Like, what are some of those new roles or personas maybe that will build on Netlify that just don't exist today in the pre-AX web? Yeah, I mean, I don't know that we figured out what to call them yet, right? But we're definitely seeing these people start building stuff without actually writing a lot of code, but having a lot of creative ideas

insight and ability to come up with stuff, right? And I'm starting, like, I've mentioned this example a couple of times before, but there's this UX designer from Niantic that's kind of become maybe like the best...

the deepest expert in building things with Bolt.new in the world right now, right? Like in this building these crazy things like a full 3D object editor in the browser with AI that he's built entirely by prompting Bolt, right? Like without writing code. And that's kind of like

mind bending that you can stretch it so far, right? Like there's these people building really cool stuff with WebGL through 3D AS entirely through prompting that before would be really expensive to build, right? Like if you wanted to build like a WebGL based first person experience in the browser, you would have to have a lot of different

like both artists and developers collaborating together and so on.

And suddenly you have these people that can start figuring out how to do that whole process on their own without writing code as such, right? But still having an understanding of the system. And that's one of these areas where I think we'll start seeing a lot of those types of developers. And we might just end up calling them developers, right? We might just end up calling them web developers, but they'll work with a very different tool set.

than the current web developers work with. Maybe vibe coders. Maybe vibe coders, yeah. Hey man, I got to get you off the hot seat, but you're not going anywhere without answering one last important question for me. So I know you've always been and are a developer at heart, but I got to think that

The Matt Billman that founded Netlify in 2014 is different from the one today who's the CEO of a company with hundreds of employees. You've raised hundreds of millions of dollars. You're a manager. You're a leader. You're recognized as an industry expert. What has changed most about Matt in the last 11 years?

Good question. I've learned more about myself and what I'm good at and what I'm not good at and so on just by doing a lot of stuff.

you learn more about your strong sides and your weak sides, hopefully through this journey, you build a lot more confidence in some areas of them, right? Like from figuring out, hey, I actually know how to do this. And you start learning how to delegate and recruit great people around you to do the things that you're maybe not great at.

But other than that, I don't know that in the end, it's still the same thing that drives me from the beginning. I've always just loved building things that other people build things with. I think that's one of the most fascinating things with building these kind of developer tools and these kind of products that enable other people to do stuff.

that you're fundamentally building like sort of these building blocks that you then see have real impact in the world through enabling other people to reach their audiences to build their products to to to ship their ideas and and and that keeps being my biggest motivation and

And even when I'm seeing this whole space emerge of this new generation of tools and so on, I see a lot of developers that meets it with a sense of fear or like it's going to take away our... I've spent so much time learning how to do this. Now this model can do it and so on. And I don't get it. My reaction is so different when I suddenly see...

One of our principal engineers sat down with his AGL kit and the AGL kit used Bolt to build like a 2D platform game with his own drawings that the characters and so on. And we're so excited, right? Like, whenever I see something lower the barrier for people to build cool stuff, I just get excited. I love it. I mean, you still have that just, you know, twinkle in your eye, you know, when you talk about the why behind the what you do.

Matt, that was brilliant. I know that you're 11 years in to Netlify at least and just getting started. So please come back sometime. We'll pick up where we left off. Anytime, Ant. This was a lot of fun. Yeah, I really enjoyed this one. Where can the audience learn more about you and Netlify? I actually launched a new blog recently.

got to figure out how to spell my name to find it. That's going to be the challenge for you all. But if you do, I'm starting to share some of my writings there and I built the blog with Bolt and Cursor and so on using these authentic tools. So that's a place. And otherwise, Twitter, Blue Sky,

LinkedIn, all of the usual platforms, you'll find me there. But I also wanted to make sure that I got back to having my own domain with my own channel in these times where I think social media is fragmenting more and more.

We'll put a link in the show notes so nobody has to memorize the spelling of your name. How about that? Sounds good. Excellent. Well, before we wrap up, thanks to Sherry Prescott, who's a dear friend of mine and a team member of Matt's who recently joined Netlify for putting us in contact. And gosh, that's all the time we have for this week on AI and the future of work. As always, I'm your host, Dan Turchin from PeopleRain. And of course, we're back next week with another fascinating guest.

♪