Dries Buytaert on AI Slop, Agency Reinvention and the Broken Publisher Deal

Table of Contents

• The Stable Triangle — and Why AI Is Breaking It
• The AI Slop Problem Threatening Open Source
• Drupal's AI Vision: Background Agents and the Context Control Center
• The Agency Reckoning: From Builders to Accountable Strategists
• AI and the DXP Stack: The End of Hard-Coded Integrations
• The Broken Publisher Deal — and What Might Fix It
• The Tailwind Labs Warning: When Your Business Model Cannot Survive AI

The Gist

• The Stable Triangle Is Breaking: AI is disrupting all three legs of Drupal's ecosystem simultaneously — the platform, the agency world, and the open source contributor community.
• AI Slop Is a Real Problem: Low-quality, AI-generated code patches are overwhelming Drupal maintainers. Buytaert's message is blunt: don't submit code you don't understand.
• Agencies Must Reinvent: The agencies that survive AI will pivot from hourly development work to strategic configuration, AI orchestration, and accountability for outcomes.
• The Publisher Deal Is Broken: AI crawlers are extracting content at scale and returning almost nothing. Only a brokered marketplace model — or genuinely irreplaceable content — offers a path forward.
• The Tailwind Labs Warning: When everything you sell can be specified and generated by AI, your business model doesn't survive. The stress test has already begun. 

Editor's note: You wanna listen to this interview in podcast form on the go? No sweat. Check out the audio-only CX Decoded version of our chat with Dries.

Dries Buytaert had just stepped off the stage at DrupalCon Chicago when he sat down with me for this episode of The Digital Experience. What followed was less a product keynote recap and more a frank assessment of an industry under pressure — one where the tools powering the web are now threatening to upend the people who build it, the agencies who sell it, and the publishers who depend on it.

Buytaert, founder of Drupal and executive chairman of Acquia, came armed with a mental model he has been refining for months. He calls it the stable triangle. And his message to the digital experience community was direct: AI is not disrupting one side of that triangle. It is disrupting all three at once.

The Stable Triangle — and Why AI Is Breaking It

For two decades, Buytaert explained, the Drupal ecosystem has rested on three interdependent legs: the platform itself, the digital agency ecosystem that brings it to market, and the open source contributor community that builds and maintains it. The central question of his DrupalCon keynote — what he has come to call the Dries Note — was what happens when AI disrupts all three simultaneously.

"AI is changing the expectations end users have from their CMS," Buytaert said. "AI is changing the business model of digital agencies because it's getting harder and harder to charge by the hour. And AI is also impacting how open source communities produce code."

The answers, he was quick to acknowledge, are not all in yet. "We don't have all the answers," he said. "We don't know exactly all the things." That kind of transparency, he noted, is a hallmark of DrupalCon keynotes — equal parts community meeting and honest accounting of what is working and what is not.

The AI Slop Problem Threatening Open Source

On the contributor side, the disruption has a name: AI slop. As AI tools lower the barrier to submitting code patches, a relatively small group of Drupal maintainers is being buried under a rising volume of AI-generated contributions — many of which are not production-ready.

"AI can generate a thousand lines of code in no time," Buytaert said. "And the human needs to go through it line by line, making sure there are no security bugs, no impact on scalability. When it's AI slop, it's frustrating to these maintainers because you're ultimately not respecting their time."

Drupal, he noted, is not a personal side project. It powers millions of websites, including some of the most mission-critical on the internet. The stakes of accepting low-quality code are high. His message to contributors was unambiguous: do not submit code you do not understand.

At the same time, Buytaert was careful to distinguish between AI slop and AI-accelerated expertise. For developers who already know what they are doing, he argued, AI is a force multiplier. "If you know what you're doing, you can use AI to go much faster — and you won't submit slop." The problem is not AI. The problem is AI in the hands of people who mistake volume for value.

Drupal's AI Vision: Background Agents and the Context Control Center

On the platform side, Buytaert's DrupalCon demos pointed toward a future where AI agents handle the optimization work that most marketing teams never get around to. The vision starts with what Drupal calls the Context Control Center — a centralized repository where marketers store brand guidelines, tone of voice, and audience definitions. Drupal's built-in agents consult it automatically when building or refining pages.

From there, the roadmap gets more ambitious. Buytaert demonstrated how Drupal AI agents can connect to Google Analytics, monitor page performance against predefined goals, and — while the marketer sleeps — diagnose underperforming content, propose improvements, and queue them for human review.

"You would literally wake up in the morning and see: here are five pages we improved, and here's why," Buytaert said. "Then you review the proposed changes and accept them or not."

The longer-term vision goes further still. A marketer running an event could hand the system a Google Doc and a goal — 300 registrations — and let the agents build the site, create social content, and iterate daily toward that outcome. "Their mission is to get 300 people to sign up for this event," Buytaert said. "They wake up every morning figuring out what they can do better to drive attendance."

The Agency Reckoning: From Builders to Accountable Strategists

Perhaps no part of the conversation was more pointed than Buytaert's assessment of the digital agency world. The question I posed was blunt: if a CMO can prototype a website with AI in an afternoon for $20 a month, why hire an agency?

Buytaert's answer was equally direct. AI commoditizes the mechanical act of building a website. What it cannot commoditize is judgment — knowing what a great site looks like, how to extract what a client actually needs (which is often different from what they say they want), and how to build something that is secure, scalable, and governed at enterprise scale.

"The typical Lovable website — you can work on it with maybe three editors," he said. "It's built for a single editor. There is no translation. There's no real permission model, no compliance, no governance." These are the gaps a CMS like Drupal fills, and the gaps an agency is positioned to configure and manage.

His pitch to the market: use AI to prototype fast, but use Drupal to build systems that last. The agency's new role is not to write the code — it is to direct the agents, configure the AI framework, manage the token economics, protect proprietary data from leaking to AI providers, and be accountable for outcomes when the agents get it wrong.

"Agents make mistakes," Buytaert said. "The accountability for security, performance, scalability, conversion rates — that remains human. End users want to pay someone for that accountability."

AI and the DXP Stack: The End of Hard-Coded Integrations

Zooming out to the broader digital experience platform landscape, Buytaert sees AI agents fundamentally changing how martech stacks are integrated. Historically, connecting a CMS to a CDP, an email platform, or a commerce system required custom development or pre-built connectors. That model, he argued, is giving way to something looser and more dynamic.

"It's no longer about connecting APIs and hand-coded integrations," Buytaert said. "It's about designing responsive, event-driven systems with orchestration tools." He cited platforms like N8N as examples of where the integration layer is heading — AI agents moving data and triggering actions across systems without a developer writing a line of glue code.

The opportunity for agencies, he argued, is to become the architects and overseers of those orchestrated systems — a role that requires more strategic thinking and less keyboard time than the agency model of the past decade.

The Broken Publisher Deal — and What Might Fix It

Speaking as a publisher navigating the same disruption his readers face, I pressed Buytaert on content discoverability in the AI era. The concern is not abstract: AI crawlers are extracting publisher content at scale, summarizing it for end users, and sending almost no traffic back. The old search bargain — publish content, get indexed, receive visitors — has broken down.

"Right now the deal is pretty broken," Buytaert said. "You get almost nothing for all the content they extract from you."

Large platforms like Reddit have negotiated direct licensing deals with AI companies worth tens of millions of dollars. Mid-sized publishers have no such leverage. Buytaert pointed to Cloudflare's emerging marketplace concept as a potential path — a brokered model where publishers could set terms for AI crawler access, similar to how programmatic advertising works today.

In the meantime, his practical advice echoed what has always separated durable publishers from disposable ones: make content that is genuinely irreplaceable. Sign-up walls may help protect it from crawlers. But the deeper answer is quality. "If you have high quality content, people will come to your site," he said. "If you have average content that 10 other sites also have, then yeah."

There is a silver lining, he noted. When AI answer engines do refer a user back to a source, that visitor tends to convert at a significantly higher rate than organic search traffic. Whether that is sufficient compensation for the volume lost is, as Buytaert put it, the question nobody has answered yet.

The Tailwind Labs Warning: When Your Business Model Cannot Survive AI

Buytaert closed with the cautionary tale he had explored in a January blog post: Tailwind Labs, an open source developer framework used across millions of websites, whose business model collapsed under AI pressure from two directions at once.

The first problem: the pre-built UI components Tailwind Labs sold by subscription could now be generated on demand by AI. The value proposition evaporated. The second problem: developer traffic to their documentation dropped sharply — by roughly 40%, I noted — because AI agents building software no longer need to send humans to read the docs.

"Everything that can be specified, AI can build," Buytaert said. "The value shifts to operating things." Tailwind Labs ultimately laid off approximately 75% of its engineering team. "It was a great product. Great people. Just a bad business model."

Acquia's model — running, scaling, securing, and hosting websites rather than building them — is, by that logic, more defensible. You cannot prompt an AI agent to operate a mission-critical website at enterprise scale. Not yet.

The lesson Buytaert drew from the Tailwind story applies well beyond open source. For martech vendors, agencies, publishers and CX practitioners alike, the question AI is now forcing is the same one every stressed business model eventually has to answer: what do you do that cannot be specified, generated, and deployed in seconds? If the answer is not clear, the stress test has already begun.