Netlify

A Netlify edge response is intermittently slow in one region.

In 2013, my best friend Chris called me from Denmark. He'd just been diagnosed with Hodgkin's lymphoma. 80%+ survival rate, but serious - radiation therapy, chemotherapy, the whole process. I flew back to see him while he was going through treatment. We started talking about what I was building - BitBalloon, this idea of self-standing front-end hosting for the decoupled web. And those conversations during his treatment are really where Netlify was born. Christian (Chris) Bach came from the content and marketing side. He could see the architectural shift I was describing. He was the one who said: "This sounds like a whole new category. If we do something together, it has to be a real category play." Those conversations led us to coin the term Jamstack. Of course, he had to get through the treatment first. And the thing that struck me was what that whole process crystallized in him. He had been comfortable - he'd built his own digital agency, sold it to a full-service firm, was Chief Digital Officer and a partner with a good salary. But coming out the other side of a brush with death, he had clarity about what mattered. He wanted to do something more impactful. So he sold his apartment in Denmark. Moved to the Bay Area with his wife and kid. Put thirty thousand euros into the company to get an investor visa. Gave up having a salary, entirely. We bootstrapped from there. Our funding was our bank accounts emptying out. I was paying myself $3,000 a month living in San Francisco. He was supporting a family with no income in one of the most expensive cities in the world. We spent a day on the lawn at Berkeley and came up with the name Netlify. Built the launch plan. Launched in March 2015, just the two of us. I think about what that process crystallized in Chris, because it's something I still see in how we make decisions at Netlify today. Chris had been forced to decide what was worth risking everything for, and then he actually went and risked everything. There's a kind of commitment that comes from having already answered the harder question about what matters, and I suspect it shapes every decision a company makes from that point forward.

the way that I'm going to do it is using a platform called Netlify, which essentially allows me just to put static information on the internet for virtually free.

I also want to deploy to Netlify.

Netlify gives you SSL, a CDN, and a drag and drop interface.

When we started Netlify, deploying a website meant provisioning servers, configuring load balancers, and managing infrastructure that had nothing to do with the thing you were actually trying to build. Serverless hosting for the web changed who got to be a builder, because it made all of that invisible. I'm seeing the same shift happen right now with AI model inference. One example: Featherless.ai just raised $20M to do serverless hosting for open-source AI models — one API, 30,000 models, no GPU management. The founder comes from the RWKV open-source community under the Linux Foundation. This is the same movie. We watched it with web, then mobile, and now AI inference is running the same script. The infrastructure starts out complex and expensive, requiring specialists. Then someone makes it disappear behind an API, and the question flips from "how do I even set this up?" to "what should I build with it?" When the answer to "how do I run an open-source model?" becomes "you just call it," the default shifts. You stop weighing which proprietary API to buy access to and start asking which open model fits what you're trying to do. The cost comparison changes, the lock-in changes, and the pool of people who can build changes. This is what I keep coming back to with the build-vs-buy shift. Every time infrastructure goes serverless, the same thing happens. The web proved it. AI is proving it again. These kind of services used to be targeted by professional developers, but in today's world where the build vs buy equation is fundamentally shifting, all teams are now the audience.

I'm excited to join Grace Gong on the Smart Venture Podcast for a live conversation Thursday about where web development is actually heading. We built Netlify in 2014 around a thesis that the web needed a fundamentally different architecture. We called it Jamstack, and for a while it played out as expected. Then AI happened, and the same shift that was supposed to accelerate composable adoption actually froze enterprise buying instead. That experience reshaped how I think about technology waves, and helped me to see AX around the corner. The better architecture doesn't always win on the timeline you expect, because adoption is a cultural question - not a technical one. What it did clarify is the next real platform shift: agents are becoming the primary users of web platforms. The question every platform now faces is whether it only exposes infrastructure to agents, or whether it designs the full experience agents need to do real work: access, context, tools, and orchestration, with humans staying on the loop. That's AX (Agent Experience). Grace and I are going to get into all of it: what I got right and wrong, what I've learned, and why designing for the agent experience, from prompt to real software running on the web, with humans staying on the loop, is what separates platforms that matter from platforms that just exist. Not the hype version, but what we're seeing from builders shipping real things on the platform every day. If you're trying to figure out what "modern web development" even means in 2026, from the GTM side or the product side, this one's for you. Link to event in the comments. Join us live. Drop your questions in the comments and I'll get to as many as I can.