Skip directly to page content
b13 GmbH

Hauptstätter Str. 59
70178 Stuttgart
Germany +49 (0) 711 460 589 70 [email protected]
TYPO3 Events Enterprise CMS

AI-ready isn’t a feature

David Steeb
RSS Feed
A black car with a sign reading "AI" is being outpaced by three orange racing cars on a light blue background with fish illustrations.

Why the most AI-ready thing about TYPO3 is the part we built years before anyone said “AI”.

Picture a content team spending eighteen months opening web pages one at a time. That was the situation my colleague Benni Mack described in his keynote at the first TYPO3 Summit North America in Atlanta this May. He had met a woman responsible for content at a large enterprise, and her team had an 18-month project running: open every page in their internal platform—thousands of them—read each one, and tag it with metadata. By hand. Eighteen months of people doing the work the system was supposed to do.

The platform happened to be SharePoint, but as Benni said, this isn’t a SharePoint story. It’s what happens when you treat content as disposable, and then realize ten years later that it’s the most valuable asset you own.

I’m starting here because right now our whole industry is being sold the opposite lesson. The pitch is: bolt some AI onto your CMS and you’re ready for the future. I want to argue the other way—using what Benni and our friends at Crowd Favorite put on stage in Atlanta—that AI-readiness has very little to do with the feature list and almost everything to do with how your content is structured. The unglamorous part. The part you can’t buy in a release note.

A speaker stands at a podium in front of a presentation slide that reads, "The tool is not enough. Take it from a software person.

The Bolt-on Trap

Benni is TYPO3’s Project Lead and Core Development Lead—the person ultimately responsible for TYPO3 on the technical side—and he has led the core development since 2014. So when he stands in front of a room and says the tool alone isn’t enough, it tends to land. His line was: “The most expensive license doesn’t make you more enterprise. The shiniest AI bolt-on doesn’t make you more strategic.”

People sometimes ask why TYPO3 doesn’t ship a big AI feature in the core. We left it out on purpose. The field changes every week. If we welded the core to whatever this month’s approach happens to be, we would be handing our users a migration project a year later—exactly the kind of thing we spend our days helping people avoid. The core’s job is to stay stable and stay out of the way.

That’s a posture, not a gap. As Benni put it, TYPO3 is “built to be a citizen in your architecture, not the dictator.” It integrates with the systems you already run—your DAM, your PIM, your search backend, your automation tools—rather than insisting on being the center of everything. AI tooling is just another integration in that picture, and a pluggable one belongs in the ecosystem, where it can move at its own speed, not bolted into a core whose support window runs up to seven years.

What “AI-Ready” Actually Means: Structure

So if it isn’t a feature, what is it?

In his keynote Benni split content into three layers, and this is the heart of it. The first layer is the content itself—the asset. The third layer is the visual presentation, which is disposable; you redesign it every few years and the content underneath doesn’t care. The interesting one is the middle layer, what he called content design: how the content is structured. Which press release belongs to which campaign. Which translation derives from which source. The metadata schema, the taxonomies, who wrote a thing, when, and who approved it.

This is the part we have been quietly hardening in TYPO3 for two decades, and it’s the same thing we mean when we talk about building core-near: staying close to a relational, well-structured content model instead of dumping everything into a couple of generic tables and sorting it out later. It is not glamorous work. It is also exactly what makes content usable by a machine.

Because once the semantics are in place, as Benni said, “you can actually do whatever you want to do with your content” — push it to any channel, hand it to another system, or feed it to a model. The structure is what makes your content portable into an AI future. Without it, you are back in the SharePoint room: people, by hand, for eighteen months, trying to reconstruct meaning that was never captured in the first place.

A speaker presents on stage, discussing the separation of content from presentation, with a slide displaying the number "1" and key points.

Your Data, Your Model

After lunch, Benni came back on stage with Karim Marucchi of Crowd Favorite for a conversation specifically about AI in TYPO3. (Crowd Favorite, for anyone who missed our earlier post from Atlanta, is the enterprise WordPress agency whose team gave us one of the sharpest outside looks at TYPO3 all day.) Karim asked the question everyone in the room actually had: how is TYPO3 preparing for the intelligent web?

Benni’s answer was that it is already prepared, and here is the part worth underlining for anyone making decisions about an enterprise stack. The companies he works with don’t pour their content into ChatGPT. They run their own, specialized AI tooling, and they feed it their own data. Or they feed structured content into their search platform and get richer results across the site. The common thread is ownership: the enterprise keeps its content and its models close, instead of shipping its most valuable asset off to someone else’s SaaS and hoping to get the value back out later.

Structured content is what makes that possible. You can only safely feed a model your own data if that data has shape—if the relationships, the provenance, and the metadata are actually there. It’s the same argument we make about avoiding lock-in, pointed at AI: own your content model, and your options stay open.

It isn’t only about content, either. Benni was clear that the same shift is helping the people who build and run these sites, that we are “about to move out of this moment of prompt-only,” and that the tooling for developers gets better, in his words, every week. For an agency, he said, you just leverage what you have always been leveraging. The foundation doesn’t change. It extends.

Two speakers sit on stage during the TYPO3 Summit North America, with a large screen displaying the event logo and aquatic visuals in the background.

The Other Reason Structure Matters

This next point has nothing to do with technology, and it’s the one that convinced me this was worth a whole post.

The web is filling up with machine-generated content—text, images, video—produced to fill pages, chase rankings, and hit quotas. Benni’s estimate, and it is hard to argue with, is that within about five years most content on most websites will be made by something that isn’t a person. Consider what that does to value. The content your humans actually wrote, edited, fact-checked, translated, argued about in meetings, and signed off on becomes the rare thing. That’s the asset your competitors can’t replicate by spinning up a model. In his words, it’s “the thing that compounds in trust and in value over decades.”

So the conclusion isn’t anti-AI. AI is a useful collaborator, and we use it. The conclusion is about what the system is for. The substrate of value is human work. Or, in Benni’s own words: “Treat your human content like it matters, because increasingly, it’s the only kind of content that does.”

A CMS that treats content as a long-lived, structured, governed asset is, almost as a side effect, the right home for that scarce human content—and the right place to point AI at it. AI-readiness and good content governance turn out to be the same project. (Benni made a related version of this argument earlier this year in The Internet Is No Longer Just for Humans—AI Bots Love Markdown, if you want to go deeper on the technical side.)

A speaker presents information about a 277-year-old premium brand, showcasing a website with design elements and project details.

It Already Runs This Way

None of this is theoretical. One example Benni showed is a brand most of you would recognize: Villeroy & Boch, making premium porcelain since 1748. Their Professional Portal serves tens of thousands of tradespeople around the world—the people who install V&B products and need installation instructions, downloads, and personal lists for their clients. We rebuilt it on TYPO3, with a design relaunch and a technical upgrade in one go, and migrated more than 10,000 pages across nine languages.

That migration is the SharePoint story with a better ending. Doing it by hand would have cost staff hundreds or thousands of hours; because the content had structure, we could write an automated migration script and let editors get on with their actual work. And the payoff of the three-layer idea shows up in what their Digital Art Manager, Jens Kelch, told us: “CMSs can get in the way of designers, but b13 freed us to deliver a user experience without limitations.” Solid structure underneath is what lets the presentation layer move freely—and what leaves the content ready for whatever comes next, including AI use cases the brand hadn’t even scoped at launch.

A speaker presents at a conference, discussing how TYPO3 enhances user experience, with a quote displayed behind them.

When This Argument Doesn’t Apply

The structure-first case isn’t universal. If your content is genuinely throwaway—a campaign microsite with a three-week life, a one-off landing page nobody will ever migrate—then careful content design is overhead you don’t need, and you should skip it. The whole argument here is for organizations whose content is a long-term asset: the ones who will still be running, translating, auditing, and building on this content in five or ten years. If that isn’t you, don’t let anyone sell you governance you won’t use. But if it is you, the structure isn’t optional, and it certainly isn’t something you bolt on at the end.

So, Are You AI-Ready?

It’s a fair question to ask your stack—but ask it of your content model, not your feature list. Can you say which translation came from which source? Do you know who approved a given page, and when? Could you hand your content to a model you control without an eighteen-month tagging project first? If yes, you are more AI-ready than most of the companies shopping for a bolt-on right now. If no, that’s the work—and it pays off whether or not you ever plug in a model.

This is the kind of thing we think about constantly at b13, and it’s a conversation we have often with enterprises weighing how to make their content ready for AI without betting the core on a trend. If that’s on your mind, get in touch—we are happy to talk it through. And if you want the wider view from Atlanta, our field report from the Summit is worth a read.

Ready to Get Started?

Are you ready to make your content work for AI?

Get in Touch

Read More Posts From Our Series Covering the North America Summit:

This post is part of our series covering the North America Summit and the insights we gained. 

Written by:

Bild von David Steeb

b13 Founder and CEO, David, is someone you’ll meet if you’re thinking about working with our agency. He is constantly learning new skills and is an influential part of our cutting-edge frontend practice.

David Steeb CEO
more from David Steeb

Related Articles