Ep. #150, The Evolution of Jamstack: An Eight-Year Journey

Brian Douglas: Welcome to another installment of Jamstack Radio. On the line we've got Matt Biilmann. Matt, what a pleasure to have you back on the podcast. How you doing?

Matt Biilmann: Good, good. It's real pleasure to to be back. It's been quite the run since episode two when I was first on.

Brian: Yeah, yeah, it's funny because you were technically the first episode recorded, but we did put you out as episode two, but yeah, I mean we started this thing eight years ago this month.

Matt: Yeah.

Brian: And it was, I joined Netlify and then the next week Heavybit had put me on an email chain of like, "Do you want to start a podcast? You have podcast experience," so maybe like you or Chris mentioned it.

Matt: Yeah.

Brian: And I was like, "Sure. Yeah, I would love to do something that was like, I've got more than 10 listeners." Which my previous podcast was like 70 downloads a week. So yeah, come a long way and like super happy for the opportunity to start this thing, but do you want to catch us up on, Jamstack today, what's Netlify doing today?

Matt: Yeah, absolutely. Yeah, so I mean I obviously at some point coined this whole "Jamstack" term, right? I had it brewing for a bit before I gave the first sort public presentation around the Jamstack at the Smashing Magazines Conference April, 2016, I guess, which was like right around the time also where you joined Netlify. Just a bit after that I guess, right?

So it's been more than eight years now of Jamstack I guess. And regularly when I coined that term, right? Like it was very, very, I think people often forget how different the world was back then in 2016 and how different the ecosystem was back then.

The way almost everyone would build any website, web app or web store was typically by picking some monolithic framework that was the backend and then kind of had some frontend added inside, right?

Like, so it would be like WordPress, that would be the whole CMS and you would have like PHP-based templates where you could put your frontend code in some symbol asset pipeline, or it would be Ruby on Rails where you would typically work in ERP templates and then there would be some kind of asset pipeline or it would be Emergent or Shopify storefront or at the enterprise scale, maybe some digital experience manager like Site Core or Adobe Experience Manager, right?

And it would always like be this coupling of the frontend layer and the backend layer. And at the same time we were sort of right in that world where like there was this emerging layer of pure frontend frameworks that sort of had ideas of how to manage components and states directly in the browser.

Like Ember was probably at the time like the biggest framework with Angular, like Ember and Angular were probably the biggest. And React was just sort of starting to emerge just as a simple view library you might pull into something like Backbone.

There was these built tools where the popular ones was like Gulp and Grunt. And then there was like this landscape of static site generators like Jekyll and Middleman and Roots and a few others that were really sort of starting to target on like this whole idea of like what if you ran some build process and just like pushed out a frontend.

And the frontend framework sort of like Ember was one of the first to introduce the idea of like an Ember CLI that could like be used outside of your backend application, right? So it's a very, very different world.

And I had started like the first predecessor to Netlify BitBalloon and then Netlify and had sort of really built conviction that the right way to architect for the web to solve a lot of the challenges of the web at the time would be to more fundamentally decouple that actual web UI frontend piece from the actual backend piece, right?

And there was all these different terms, like there was a community forming around static site generators, there was a community forming around progressive web apps and single page applications. And there was some general community around the API economy and the ideal that a lot of the backend functions would turn into APIs that you reuse.

At the time when I coined the term, Jamstack, Contentful was like a tiny startup had raised a seed round in Berlin, right? And where mostly in their communications focusing on like the need for hit list content because you wanted to support mobile applications, right?

Like and in general it was sort of the whole age of mobile applications where when we were out raising the first round of funding for Netlify before you joined, right? Like a lot of the VCs we talked to were just kind of like a little flabbergasted, like why would you do anything with the web? Isn't everything just like mobile and instant now, right?

Like, so it was really, really different world and we felt that to make this like necessary shift happen in the web ecosystem and sort of pull the frontend out of the backend and make it its own thing and start building with like built tools and frameworks dedicated to that, having like the frontend repository be like its own repository in GitHub with its own release cycle and so on.

We needed like a common terminology to rally around, right? Like there needed to be a word for that. Like we couldn't go and explain everything I just explained every time we wanted to mention this, right? Like we kind of needed to just be able to say this thing, right?

Like so we were brainstorming like names and so on and at some point over a glass of wine, a friend of mine, like when we were talking about it said like it should be something like the stack has changed, right? Like there's been like the lamp stack.

What about like wherein I know like a Jamstack, and then it was like jam, jam that could be like JavaScript API and markup. And then at a point we're like, "Okay, that's good enough." And we started talking to people in the industry and the rest is kind of history.

I gave that presentation in San Francisco in 2016 and that sort of kickstarted the whole, the one movement. And the core concept kind of work that by giving a name to this architectural shift, people could rally around it and do things with it in a different way than when it was just a bunch of different technologies without any sort of coherent division.

Brian: Yeah, yeah, and it's hard to go take yourself back to what the before times. Like I've got a 10-year-old and like they're watching shows where they think of the 90s as like an old archaic like decade. Like they think the 80s is like so far away from them.

And like to me I'm like, "Oh, that's offensive." But like what you think about 10 years ago, like it was mobile, like took over, like everyone had a mobile phone, it got saturated. And that's a real question that a VC would've asked 10 years ago.

Matt: Yeah.

Brian: But I've been watching a lot of these founder stories about like Henry Ford and like think about how a car was super expensive to build, but instead he built the assembly line so you only have to focus on one component.

Matt: Yeah.

Brian: And like you just become an expert on the frontend or the backend or the carburetor. And then you put it all together. And that's what I always see the Jamstack as that opportunity of get really good at the one piece of stack then you can then focus on the next piece and then we get better together. So like my question to you is like where do you see that's like taken us today? How that's like evolved now? The web.

Matt: Yeah, yeah, I mean I think the first thing is that the Jamstack kind of in many ways took over, right? If I think back to that whole eight year period, right? Like if you think about all of these tools like the CMSs, the e-commerce platforms, the platforms that used to be the monolith we were building with before this like decoupling, right?

Like I can't name a single one that's like emerged in those eight years like an e-commerce platform or website platform, CMS platform that's not hit lists, right? And that's not decoded, right? Like, and now we are sort of seeing even the traditional players like the big DXPs like Sitecore and Optimize and so on, build out hit list solutions and we've just seen that whole thing shift.

And when you think about the frontend piece, what we are seeing now is just like everybody is reaching for like extra frontend technology to build their frontend with, right? Like there's exceptions to the rule, right?

Like, but generally people like are thinking about okay, we need like some kind of frontend framework, frontend toolkit, it build system optimized for that. We need a whole workflow optimized for that, right? Like, so in a certain way the Jamstack just became really big just right?

And just became kind of synonym with like modern web, right? And this like original requirement we have to decouple the frontend from the staggered tools and have our own tool chain and so on. It's maybe more of like, just keep doing what you're doing already, right?

Then at the same time, I think as it happens with all of these cycles, there was also a part of what we did at Netlify and how we thought about the Jamstack and the technologies we were pushing that were a lot about pushing a simplicity of the stack also, right?

And like really thinking what's the simplest tool set that you can easily reason about and work with and so on. In that original presentation I had this whole sort of diagram of like how the current web architecture was looking like with all the different layers of CDN and servers and databases and caches and everything that went into just like operating a website of scale, right?

And originally I think we were pushing hard with the Jamstack push to also get away from that level of complexity into a mentally simpler model or websites often centered around like more a static view of the world where you could build as much as possible upfront and just distribute it on the itch. And then from there on you kind of just had to worry about what happened in the browser and sanctuary simple.

For single page applications, just sort of really say, okay, just build an app in the browser and talk to different APIs. And it's very simple. And in that regard, I think we've sort of seen the curve of complexity in the modern frontend system go up and up, where today the most popular framework is Next.js, right?

I can sort of the most widespread in usage also on Netlify, right? And there, when I go and look at just like the docs for caching on Next.js, the illustration that have it there like looks almost exactly as the illustration I had of like all the complexity in like the the pre-Jamstack monolithic web architecture, right?

And I think now I'm sensing from a lot of people that feeling of like, okay, how do we bring back some of the simplicity again, right? Like what's the path to unwind some of the complexity. So that's one part of it.

And then I think we are also in this like slightly interesting moment where I think there's a lot of new things about to come that we haven't seen yet, right? Like where in many ways it's a while since I've seen sort of a new website that like really, really wowed me, right?

Like from the technical perspective, right? Like with the current frameworks, like a lot of them can do really cool things and we take them for granted now, right? Like I think that also means we are about to experience like the next shift in people coming up with some new really cool things that will lift the expectations for what we can do.

And some of those things will probably be tied to what we can do with large language models and so on and how we can realize different kind of experiences. So those are are two moments I pay a lot of attention to.

One, the things we are doing now and the things that work well, how can we do them in a simpler way with less complexity? And two, what are the next thing that might require some more complexity, but where it will really feel worth it because it lifts the actual experience we're building on the web?

Brian: Yeah, that's amazing too as well, because like again, I think you're spot on about like the complexity of what we have today. It also feels like the complexity we had like eight years ago.

But I did want to actually ask the question or really talk about the new partnership you all just announced with Astro, because I find that extremely fascinating that that simplicity like, and like Astro basically doing enough of the work, but getting out of the way is what I love about the Astro project.

But I'm curious like what do you see? Do you see that help ushering in the next phase?

Matt: Totally, totally, right? Like Astro I've been excited about for that reason, right? I've given some talks where I've kind of pointed out like sort of through real world options right now that I feel simplifies in different ways, right?

And I think Astro has really sort of taken the... Most of the sites and e-commerce stores we build are very sort of content-centric. And they've sort of gone back to like the content-based pages model and said like, how do we go more back to just like the simple static side generator feel of like, for each of those pages, here's like the HTML that should generate.

And if we don't need it, let's not like turn all of that HTML into a single page application and everything, right? Like let's just keep it really straightforward, lean into the actual web platform, just like let the browser do what it's really good at. Use technologies like View Transitions to make like navigation between pages feel even smoother and so on.

And then they've embraced this idea of an island architecture where you then go in and say, "Okay, we have like this plain HTML page, right?" And then we probably have specific parts of that page that needs interactivity.

Like we might have like a shopping card or a button to add to card or like a price widget or something. There can be a bunch of different things, right? Like, but instead of like turning the whole page into essentially a single page application after it loads just for those bits of interactivity, let's have a way to just like pull in modern frontend components for those elements.

And then the framework will handle all this tricky logic behind the scenes of turning those into small chunks of bundles that gets loaded, right? And gets like hydrated in the page in the right place and show loaders meanwhile and so on, right? Like it's like a really simple mental model, right?

Like, and the way they sort of leaned into like this classic markdown with front matter mental model that we had all the way back from Jekyll and then used it to really say that front matter is sort of like your server side.

Like here we load stuff and do data fetching and then the stuff below it kind of just like your HTML page where here we build out the server-side rendering.

That feels like a very simple and intuitive approach and it works really well and like makes it easy to build sites that don't ship unneeded JavaScript that like very high performance that just leans in on the actual web platform itself and then opens up even to say like for those interactive components, you can kind of just pick and choose what components library you want to use, right?

Like you want to use React, just pop in React. You want to use View just, plop in View. You want to use Solid or Reactor, well, you can just pull in components of those counts and it just works.

So Astro has been very excited for that reason, right? Like, and I've been building a lot of just sort of my prototypes and I'll be projects with it and it just clicks really well, right? I can feel very easy to reason about what's going on and what's it doing.

Brian: Right.

Matt: And then the other one I've called out a couple of times is Remix that also started its life as sort of a venture back startup from the founders of React Router, then they got acquired by Shopify.

And it's kind of obvious that Shopify is really seeing them as like an core open source investment in like web infrastructure in the same way as they've been investing a lot in Ruby on Rails. And there's some similarities of the mentality of Remix and Ruby on Rails. It's probably also attracted Shopify to them, right?

Like, but they've kind of gone the other way and said like, "Hey, if instead of all these different rendering modes and something is static and something is pre-rendered and something is ISI and something. Let's just lead into like server-side rendering plus caching."

And I will say that realizing server-side rendering with caching is trickier than what most people think and we are still not there, but it's a lot, lot easier than it was eight years ago because of platforms like Netlify, right?

Like where you essentially have this guarantee of like a serverless rendering layer plus a CDN caching layer that have like very clear conventions for when purchase happens and how to refresh the cache and there's a CDN cache in front that you can rely on and so on, right?

Like, so it also becomes a really interesting model that then says like if you are building something where there's so much dynamic stuff happening that the more content-centric approach of like, just think about a page that has a set of content that builds out it's maybe not the right one and you need more of an application approach where you think about a router that like defines what data to load and then you navigate within it and so on.

I think that's allowed Remix to also say like, "Okay, if we just do that, we can do something that feels a lot simpler than like the world of every single rendering mode and complexity, right?"

And then because they're no longer a venture funded startup, they've also been able to just say, "Hey, we don't need to sort of artificially expand the scope of the framework to capture value, right?

Like we can just turn Remix essentially into a V-Plugin and then even sort of merch Remix and React Router and just like really think about like how simple can we make the architecture, right? Like, so those are two of the ones I see that even customers at scale today can go out and use, right? Like we're using Astro for netlify.com ourself, right?

Like so we know it works and it's like robust. And we even have some of our largest enterprise customers building really, really complex sites with Astro, right? And we've seen it works.

And a similar way in like Shopify's own dashboard is built with Remix now, right? Like this sort of real backing behind it there. Then there's a bunch of other stuff I could talk to that's probably like the next generation that might not be my advice to like a big enterprise, go pick this up and it'll be ready for you right now. But that can really move the needle in the future.

Brian: Yeah, and so that actually leads to like a really good segue to what's next because you're talking to customers, you're talking to the community, you've got a lot of legacy of Netlify working in this space.

But what is the feedback on, I'm a large enterprise, I want be nimble, I want to be able to ship stuff. So how are they approaching this problem today?

Matt: Yeah, I mean when we talk to the larger enterprises, they're this sort of, they are also thinking beyond the frontend layer of it, right? Like, and they'll typically talk more about the whole concept of how to make their architecture composable, right?

Like they've kind of in every part of the architecture, they've kind of been bitten when they've lent too much into like one big monolithic vendor that covered a very big scope going from often vendors that would give them like a digital experience manager and they would do everything from like, this is how the website gets rendered and built and this is how content is managed and this is how e-commerce is done and this is how search is done, this is how personalization is done and experimentation is done.

And they, like in those cases, you just inherently end up in a situation where a lot of those individual concerns fall behind and you can't just go pick the new thing that's now best in class because you're tied into that whole monolith, right? So a lot of large companies are really thinking of how do they split this up and how do they make their architecture compostable, so they can pick their content provider independently from their frontend framework, right?

Like so they can pick their e-commerce provider independently from their search engine, and so they can pick their personalization tooling independently from, again, from their content or from there, right? Like and also so they're ready for a future with a lot of change coming.

Probably I think everybody is looking to everything that's happening in generative AI in large language models and seeing like, "Hey, we don't know exactly what will happen because of this, but it's going to be a lot of change we have to keep up with and it's going to change our customer's expectations to what our web properties should do for them a lot."

So if we lock ourself into like one monolithic system there, we could really get into a bad position where we can keep up with that change fast enough because we picked the wrong vendor and we can't swap any piece in it, really, right?

So I think a lot of the companies are seeing like how do we get from where we are, so that world where we can easier swap in and out things.

And obviously as Netlify one of our big messages is like, you've got to start by making sure that the layer you have for building, deploying and operating the actual web experience is decoupled from your other systems, right?

Like, because that web experience is what's actually going to change the fastest and that's where the change actually makes a difference, right? Like that's where you to see it. And once you have that, you will be in a much better place to plug in different services.

And I've sort of really put us on a course to be really clear around like our philosophy of primitive soil frameworks. Netlify's role is to really think through what are the right core platform primitives that will last longer than any of the individual pieces that we can build a really great developer experience around that we can easily map any modern frontend framework tool, right?

Like so we can guarantee that you can run whatever framework delivers the most value right now for your customers and then really help companies take care of all the automation and operation that goes into actually like shipping frequently and keeping your site live and so on.

And allow the the creative team and the developers to focus on like, "Okay, what components should we pick? What framework should we pick? What technology do you want to work with? And how do we do the creative work?"

Brian: Excellent, yeah, so I had you on the podcast and I reached out to you specifically because after eight years and sort of burying the lead at this point, like this will be the final episode of Jamstack Radio.

Matt: Yeah.

Brian: And it's a emotional decision to come to because you know when you have a product or a thing that you've been building for so long and having all these conversations that one of the feedback I got from folks is while doing this podcast is it seems like you have everyone on this thing.

And what I really pride myself on is like being able to find folks early days before they become big. So like Prisma, got to talk to them. Gatsby, talk to them, early days. So quite a few projects and founders got to have those conversations and including yourself.

Matt: Yeah.

Brian: Got to talk to you on this podcast pretty early before Netlify and it's like it's massive valuation of what it is today.

Matt: Totally.

Brian: Like early days.

Matt: So what were you, when your joined were we like six people or something?

Brian: Yeah, yeah. That was less than five, yeah.

Matt: Yeah.

Brian: But yeah, small team in the Dogpatch, which even today, like going to the Dogpatch and like now we got the Warriors playing out there. Steph Curry just bought a building across the street from the old Netlify office. And so he is going to set up his VC fund, like right there in the Dogpatch, YC is now in the Dogpatch.

Matt: Yeah.

Brian: So you and Chris are this early on the space of a few different things. Of that neighborhood, but also about the web. So it's been amazing just to see this ride and see like where it's blossomed. But I'd be remiss to not mention Composable Web, the event that's happening here in October in San Francisco.

Matt: Yeah.

Brian: What are the like high-level themes that you see that are happening at that event? Give us a reason why we should be buying a ticket right now.

Matt: Yeah, I mean, Compose in San Francisco, it'll be October 3rd, it'll be a broad developer event. We're focused on the modern web on how to unlock the power of the modern web that's sort of core to our vision, right? Our mission, right? Like how do we help developers unlock the power of the web?

We'll have people from the different aspects of frameworks and frontend development. We'll obviously also have people in the space of AI in the terms of like how is AI going to impact what we built for the web and how the web looks like.

And we are going to have customers and partners share sort of real world stories of things they've built. I always think like sometimes ends up being some of my favorite part of every one of these conferences, just seeing actual builders coming and show like, this was how we actually ran this kind of thing at scale, right?

And then obviously as Netlify we are pretty excited to show up with some really cool product launches that the team is working hard on now. But most of all with all of these conferences, it's all about like the people and getting together as a community, right?

Like, and I think what's always been my favorite part of the best conferences we've run have typically been that we sit sort of in a pretty neutral space between like all the different frameworks, all the different CMSs and commerce platforms and emerging APIs and toolkits and databases and so on, right?

Like, so it's like an interesting place for like a broader product, the modern web community to come together and interact and hear from each other, learn and grow together.

Brian: Excellent. Yeah, so, folks check out Compose. And, yeah, buy a ticket and we'll see you there.

Matt: Yeah, absolutely.

Brian: Cool, so I want to transition us to the final Jam picks. These are things that we're jamming on, we'll always be jamming.

There's opportunity for us to just to share our mind of like what we're working on, what's exciting us. So I'll go first if you don't mind. I've got a couple things that I'm jamming on. I've been working on open source for the past almost two years at this point, and we're about to ship a new feature.

So at the time this episode goes out, it's going to be called the OSCR, so the open source contributor rating. And what we've been doing is we've been crunching GitHub data and I've been doing a lot of data science. I've learned how to do this.

Like back last Christmas I spent two weeks crunching numbers, learning how to use a SQL as a data scientist and then using AI to like interpret and predict things that's very similar to like a FICO score. So if you think of like your credit rating, think of like your open source contributor rating, very similar.

So, we take not just code contributions, but we're also taking comments and the ability to instigate conversation in projects. So it's going to be a lot of fun the ship. So I look forward to like maybe the end of the year we'll do like the Oscars, we sort of announce some of the top developers within the space.

But yeah, would love everyone who's listening to check it out and find me in other places as I wind this down. The other thing I'll mention is Resend. We've been using Resend since the beginning of React email and it is amazing product like being able to generate your emails.

Like my earliest experience is actually sending emails at Netlify using Postmark. And that was a good experience. With Postmark, but I always felt like the technology never really kept up the speed. So Resend gives me something that I know how to code in and then we've been using it so for so long.

So like I would love everyone to check out Resend, if you have the time. Matt, do you have picks for us?

Matt: Oh yeah, I mean obviously a lot of the things that I'm working on prototyping internally, and so you've got to come see it. Compost show preferred, right? I'll try to keep quiet about them until then, but some of the things that I'm digging into and so on.

One is just like playing around with all their caching primitives. We kind of like build out at Netlify over the last year actually, right? Like again, sort of really looking into, if we really embrace that we have this system where the frontend code and the CDN lives together and can be programmed together, right?

Like how can we build the right patterns to make it caching really programmable and really powerful, especially like they working on building some material, really showing off like how we can use techniques like Netlify-Vary and durable caching to do things with CDN caching that would just have been really, really hot before, like easily distinguishing between like logged in and non logged in users and making certain cache tests.

It's like permanent instead of just transitory and so on, right? Like, so that scenario I'm playing around with and that I would encourage people to take a look at and dig into. Another thing I'm just like keen about is a bunch of the developments in local first.

Brian: Yeah.

Matt: When I look at sort of where I sometimes see something that really impresses me in modern apps, right? Like, that's often key to it, right? Like this feeling of like, hey, this feels almost magic.

So a few of the things there that I'm excited about, like Fireproof Storage is a project by J Chris Anderson that did the Couchbase and so on before and had a lot of experience. And Zero is a project from the team that did Replicache. That's sort of really, really fascinating and exciting.

And you mentioned Prismic before, where Johannes is from there is is working on live store now, it's like a core sink and it just feels like one of these spaces where there's so much cool stuff going on that following those projects and playing around with them. We'll probably give some insight into what's going to happen in the future of web development.

Brian: Excellent. Yeah, there's a lot. At the look of first thing I get really excited about because I love the concept. I was going to ask you about the Arc browser in our conversation. I completely missed that opportunity to jump in, but I feel like there are new paradigms even on the web that are making things a little more interesting. And yeah, I always been on the web for sure.

Matt: Yeah, and at the same time, I think a lot of those things like will also mean like similar to when Netlify got started in this mobile revolution in which we're like a little counterintuitive like why do we.

I do think we'll go into little, we kind of in the stages of early stages of that kind of face again, where the AI platforms will mean that there's going to be a clear new battle of the platform, right?

Like I think everything that's based around this idea of like chat-driven interactions and so on, they'll actually tell to be a pull away from the web and into different closed platforms like into tools like ChatGPT or Messenger or WhatsApp, or what X might turn into and so on, right?

Like there's kind of like a blueprint for how to start building new platforms that are not web and that just kind of like owns the whole lifecycle of the user interaction there, right?

I think we are also sort of right in that moment again where we're really investing in the open web and really figuring out like how do we make web experiences so powerful and compelling that the web is worth it.

It's going to be really important again, right? Like I stole this phrase from Tobias, the CEO of of Shopify, right? Like, but had great quote where he said like, "If anyone released the web browser as an app today, it wouldn't be accepted in any of the App Stores anywhere, right?" Like, it's just like way too powerful, right? It's a whole platform, right? Like, but we have it.

Brian: Yeah.

Matt: We just got to keep in investing in that layer and keep it relevant because that's the open platform we have.

Brian: Yeah, that's really true. That's actually a great quote too as well. Honestly I could probably talk to you for like hours about this, but I look forward to talking to you more in the future. Just not on the podcast.

And, folks, really appreciate you listened to us for eight years. Find me on other platforms. You'll find me, bdougieYO on the X, I should say. And you'll find me, bdougieYO everywhere else. And you can also find Matt as well. So, keep spreading the Jam.