Mountain Camp 2027

AI Needed to Happen to Make Me Fall in Love with Drupal: And why the next thing it will change is how we buy software

One-liner: I spent 14 years beating Drupal in tenders. AI changed that. Next, it will change the tenders themselves.
Short abstract:
For 14 years, I sold WordPress and competed against Drupal in tenders. I've sold hundreds of WordPress sites. I know exactly how Drupal gets beaten, and I always knew the arguments didn't really hold up. AI changed the rules in a year. Drupal's hardest sells became its sharpest differentiators, with digital sovereignty leading the way.
But the bigger change is still coming. When AI can build working software in minutes, writing a 200-page specification before anyone has seen anything stops making sense. This talk shows how AI-generated Proofs of Direction keep humans in the loop, how they're already changing private sector projects, and why public tendering is next.


I spent years selling WordPress at Miltton, and before that as an entrepreneur, running a WordPress-based company. I know why WordPress won: easier, cheaper, better liked. I also know what it lacked. Content with real structure. Fine-grained permissions. Many languages without extra plugins. Drupal has all of that built in. And its home, drupal.org, is run by a non-profit.
For 14 years, I competed against Drupal in tenders, and often won. I know exactly how Drupal gets beaten: "too expensive", "too complex", "hard to find developers", "nobody likes using it". I also knew, deep down, that most of it didn't hold up. In this talk, I'll show the Drupal community what it looks like through a competitor's eyes, and why AI has made those arguments fall apart.
Within one year, the conversation flipped. Can AI understand our content? Can we trust it? Who owns our data, and who controls our AI? Every question pointed to Drupal. Structured content became AI fuel. Security became a precondition. Digital sovereignty went from open source idealism to a board-level requirement.
AI didn't make Drupal better. It made Drupal obvious.
But I'll also say what insiders sometimes don't. Drupal still has a reputation to outgrow, and knowing why the arguments fail isn't enough. The real change is in how we work: when you can show clients working software before they buy anything, the old way of buying software stops making sense.

Code is no longer the bottleneck. Agreement is.

AI can generate a clickable prototype during one afternoon. Most projects don't fail because teams can't build. They fail because nobody agreed what to build, and everyone found out too late.
So we stopped selling projects and started offering direction. From the first workshop, stakeholders work with clickable, AI-generated software instead of documents. Proofs of Direction, not proofs of concept. Not "can we build it?" but "should we?" Feedback in minutes, not weeks.
Nobody defends a design they didn't hand-make, so people stop arguing and start solving. Then we freeze the scope, price it precisely, and build, with AI proposing and humans deciding at every step. Drupal can carry it from prototype to production: Drupal CMS, recipes, Canvas, and AI features designed on the same principle. Or then it's done by using Drupal and other AI tools.
At the University of Helsinki, this looks like a product owner who At HelsinkiUni Help, night jobs check the service, and the owner reads one message each morning. Then people decide what to fix. Tomi Mikola's technical talk shows how.The system doesn't replace people. It makes the institution smarter about its own users.

What gets lost after the workshop

Fast prototypes bring a new problem. When AI can make ten versions in a day, a month later nobody remembers which one was agreed, who agreed it, or why. The understanding is gone, and the next person only sees the result.

So we are piloting the memory for this way of working, on Drupal. Every prototype version, every reason, every decision, and the links between them, in one workspace. People approve one exact version, never a link that keeps changing. AI prepares and drafts, using only what that person is allowed to see. It never decides that people agree. Drupal isn't there by accident: structure, permissions, versions and workflows are what it does best. And the part that keeps decisions honest, approving one exact version, goes back to Drupal.

By Davos, it has run months in a real client project, round after round. On stage I'll ask someone who has never seen that project: why is this feature in the first release? The workspace has to answer in a minute, with its sources. Then I'll show the numbers against the same way of working without the tool. If the software doesn't beat the method, I'll say so, and we stop building it.

And this, this is powered by Drupal. The AI module, the ECA module, the Tool API module, the MCP Server module, and many more.

The next five years: the end of the tender as we know it

Today's public tender asks organisations to describe, in detail, a system nobody has seen, then asks vendors to price that guess. I've priced that guess many times. It was the best tool we had. Now AI makes a better one possible.

In two years, I believe the best tenders will buy direction before they buy delivery. Buyers will validate needs with working prototypes before procurement starts. Vendors will compete on how well they understand the problem, not on how cheaply they can promise to solve a document. Specifications will be living software, not frozen paper. And platforms will be chosen on sovereignty and structure, because AI will make lock-in visible and expensive.

The legal tools for much of this already exist in European procurement. What's been missing is a way of working that makes them practical. The private sector is already proving the playbook. The public sector can take it further, because it has the most to gain from keeping humans in the loop.
This is a talk from someone who has sold on both sides of the CMS divide, in both private and public sectors, about what AI has already changed, and what it's about to.

Key takeaways

  • How Drupal gets beaten in tenders, seen from the competitor's side, and why those arguments no longer hold.
  • Why AI turned Drupal's toughest sells into its strongest differentiators, with digital sovereignty leading the way, and what Drupal still needs to fix.
  • Why agreement, not code, is the new bottleneck, and how AI-generated Proofs of Direction solve it with humans firmly in the loop.
  • How the private sector already works this way, and how the public sector can follow.
  • A concrete five-year view of how AI will reshape tendering, procurement and the role of platforms like Drupal.

Who it's for

Agency leaders, sellers, procurement professionals, public sector decision-makers and product owners. And anyone who has ever had to explain why Drupal is worth it.

Ulla Koho

Ulla Koho spent 14 years competing against Drupal before switching sides. She is Chief Commercial Officer at Mearra (formerly Wunder), a digital agency working with Drupal in Finland, Estonia and Latvia. Before that, as Director of Business Design at Miltton, and as an entrepreneur running her own WordPress-based company, she sold WordPress solutions and learned first-hand why WordPress won, what it lacked, and how organisations actually decide what to buy.

She has worked with both public and private sector clients, and spends most of her time helping them figure out what to build before anyone builds it. In 2026 she spoke at Drupal Dev Days Athens https://www.youtube.com/watch?v=D27gGn2kwzA , DrupalCamp Helsinki and DrupalCon Europe youtube.com/watch?v=Jw-aygNKrFw&list=PLKmB0WkfSd4U&index=16.

Converted to Drupal by AI, and not afraid to say where it still needs to improve.