fidzu : web

Missed the Driesnote? You can watch it here.

Drupal, the open source content management platform that runs some of the most demanding websites on the planet, turned 25 in January. But while the community is celebrating what is a remarkable milestone for any open source project, it is actively strengthening its foundations to lead in the AI era and looking ahead to a future it intends to shape.

This week at DrupalCon Chicago, Drupal's creator Dries Buytaert delivered his annual keynote, the DriesNote, and it was one of the more honest talks you'll hear at a tech conference. A clear-eyed look at what's working, what's under pressure, and what the plan actually is.

AI is Disrupting Everything, But Deep Expertise is Irreplaceable

For more than two decades, the Drupal ecosystem has rested on three things: the platform itself, the agencies that build with it, and the community that maintains it. That triangle has survived waves of new technology and constant change. It's been remarkably resilient.

But what happens when AI disrupts all three sides at once? When anyone can spin up a decent-looking site in fifteen minutes, what does that do to the people who've spent years building something better? That's what is happening at the moment, as the world is being flooded with AI-generated "average". Average content, average code, average websites - average is easier to attain than ever.

What it means is that the only thing that will actually matter, to customers, to organisations, to the people trying to build something lasting on the open web, is genuine, hard-won, deep expertise.

What AI Actually Can't Do

Here's something worth understanding, because it gets lost in the noise.

AI can generate a beautiful website in about fifteen minutes. Tools like Lovable and Replit are genuinely impressive. You give them a prompt, they give you something that looks polished and professional. It feels like magic.

But a prototype is not a production system.

The moment you need structured content that editors can actually update, workflows that a real team can follow, permissions, governance, security, accessibility, multilingual support, compliance... you're not building a website anymore. You're building a system. And building systems is exactly what Drupal has excelled at for 25 years.

The demo at DrupalCon made this tangible. A beautiful event site built in Lovable in minutes, then migrated into Drupal CMS using AI coding tools, where the hard-coded layout became structured, reusable, editable content. Same visual ambition. Completely different foundation.

The pitch is simple: AI gets you to visual ambition fast. Drupal makes that ambition durable.

What's Actually Shipping

This isn't a vision talk. Things are being built and released.

DrupalCMS 2.1 landed at DrupalCon, built on top of Drupal Core 11.3. Over the last 18 months, core database and cache utilization have roughly halved, meaning every Drupal site in the world gets faster when it upgrades. That's not a minor thing. That's the compounding benefit of a serious engineering community.

Site templates and a marketplace are now live at marketplace.drupal.org, with more than ten purpose-built templates covering nonprofits, education, healthcare, events, government, and SaaS, built by agencies that understand those sectors. Free and premium options, with direct access to the people who made them if you need help.

Canvas, Drupal's new page-building layer, lets teams create and customise pages at speed without sacrificing the structured content underneath.

The Context Control Centre is a system for storing and managing your organisation's institutional knowledge (brand guidelines, content strategy, audience personas, live analytics) and it's moving from prototype to production. The idea is that AI tools are only as good as the context they're given. Without it, you get the average of the internet. With it, you get something that actually knows your brand.

And in the AI layer itself, a demo showed what it looks like when a marketer can drop a raw content brief into Drupal, have the system read it, load the right brand and strategy context, ask clarifying questions, and generate a production-ready page, with proper cross-linking, structured data for AI search engines, and an accessibility check built in.

That's not a concept. That's a demo running on real code.

One Developer, Six Weeks, 90,000 Lines of Code

The most striking moment of the keynote was a contribution from Jurgen Haas, one of the Drupal community's most experienced developers. He builds ECA, Drupal's automation engine, running on thousands of production sites.

Three years ago, he knew what ECA needed. He knew how to build it. He never had the time.

Six weeks ago, he started. With AI as a collaborator, handling scaffolding, generating tests, refactoring code, he shipped a completely rebuilt workflow editor: a new visual interface, built-in debugging and replay, in-context automation for non-technical users. 90,000 lines of code. Full test coverage. One person.

"This is what one Drupal developer can build in six weeks," he said. "Imagine what all of us can build next."

The key detail: Jurgen could explain every line. He could defend the architecture. He owned what he built. AI removed friction. It didn't replace expertise.

The Harder Conversation

Not everything in the keynote was product news.

Dries was honest about the pressure on Drupal agencies. When AI commoditises production, and it is, the business models that agencies have built over years start to look shaky. An agency leader named Aidan Foster, seventeen years into running a Drupal shop, described the feeling plainly: "AI had converted making things into a commodity. That shook the foundations I had spent 17 years building."

But Aidan's conclusion was interesting. The bottleneck isn't production anymore. It's creativity, strategy, and judgement. If you use AI without asking the hard questions, who are we, who are our audience, what makes us different, you get the boring average. The agencies that will win are the ones that get good at encoding expertise, not just delivering outputs.

There's also a challenge for the community itself. AI lowers the barrier to contribute code, which sounds good, until you realise the burden of reviewing that code falls on the same small group of maintainers. And when people use AI to skip the deep learning that used to come from contributing, the community gets shallower. A shallow community can't maintain what's been built.

Dries' response was a new mantra:never submit code you don't understand. It doesn't matter what tools you used to write it. If you submit it, you own it.

The Bet Worth Making

Twenty years ago, Dries was a bedroom inventor who collapsed from stress on a street in Belgium. He had a choice: take a safe job, walk away from the thing he'd built, or ask for help and become a deliberate leader.

He made the harder choice. The community that grew up around that choice is why Drupal is still here, still relevant, still running critical infrastructure for organisations around the world.

Now there's another crossroads. AI is both the flood and the drainage system. It destabilises the foundations and it can help rebuild them stronger.

Twenty-five years of Drupal is twenty-five years of expertise built patch by patch, merge request by merge request. A community that showed up not because it had to, but because it cared. That's not a liability in the age of AI. That's exactly what this moment needs.

DrupalCon Chicago runs through this week. The marketplace is live at marketplace.drupal.org. The Context Control Centre is approaching production. The Drupal AI initiative is moving fast.