Inside AI is building what our partners asked for

By Christoph Breidert, Product Lead, Inside AI

A year into the Drupal AI Initiative, AI inside of Drupal has a clear goal for the months ahead. At DrupalCon Rotterdam at the end of September, we want to show a single Drupal site where AI Search, an AI chatbot, AI content review, and AI translation all work together on real, multilingual content, with observability and security running underneath. The idea is simple: rather than describe these features one by one, let people see them working together on one site.

This post is about how we plan to get there, what our partners told us to build, and the team we are putting together to build it.

Two streams, working side by side

If you are new to the split, the Drupal AI Initiative now runs in two streams. “Inside AI” is AI inside Drupal, for the people using it. “Outside AI” is AI outside Drupal, acting on it through external agents. The simplest way to hold them apart is this: with Inside AI a person uses Drupal and Drupal uses AI to help; with Outside AI a person uses an agent and the agent uses Drupal. We introduced the two streams in an earlier post, and the distributed leadership structure behind them shortly after. This post is about the AI functionality we are building inside Drupal. Outside AI, led by Scott Falconer, will get its own update, and Dries Buytaert has already written about why that stream matters.

We asked the Drupal AI Partners what to build

Here is something we have not shared publicly before. We asked our Drupal AI Partners, the organizations that fund and staff this initiative, which features they most want us to build inside Drupal. The results were clear.

1. AI search
2. AI content reviews
3. AI translation
4. Chat-driven content editing
5. AI-powered bulk content updates

Search was the standout. The next four clustered closely together, which tells us there is no single second priority so much as a group of capabilities people want in roughly equal measure. This shapes what we prioritize. Drupal AI is funded by our partners, therefore we build what they asked for, in the order they asked for it, against the public 2026 roadmap already in flight.

Two of these deserve a word. We have already been building chat-driven content editing in Canvas AI, and that work continues. Bulk content updates we are not tackling as a standalone feature yet, but it belongs naturally with content review. Once you can review a large body of content with AI, the obvious next step is letting AI help you act on what it found, so for now it lives within the content review work rather than as a separate capability.

Shippable features, proven in the demo

What matters most to anyone building with Drupal AI is shippable features. The demo is how we prove they are ready, by showing each one working on a real site rather than only in isolation. That is why the demo is the most important thing we are building this year: it turns a list of capabilities into features our partners can put in front of their own clients. The two are not separate efforts. The recipes and configurations that make a feature work in the demo are largely the same ones that make it adoptable on a real project, so building the demo is how shippability becomes visible.

This is also why we are putting real effort into engaging, realistic demo content, a site with enough depth that the AI has something meaningful to search, review, and translate. The goal is for anyone to try it out with a single click and see for themselves how a fully AI-powered CMS behaves.
 

AI Search in Drupal 

Search is a natural place to begin, since it was the most requested feature. Picture a search that returns an AI-generated summary, the sources behind it, and the ranked results below, the way modern web search now works, but over your own site's content.

Search sits alongside the other capabilities, and each carries real weight in the demo. A chatbot drawing on the same retrieval foundation. Content review that scores a page against criteria like brand, legal, and reading level and suggests concrete fixes. Translation that moves content across languages with quality worth presenting. And underneath it all, the observability and security layers that make the experience credible for production rather than a set of features that only hold up one at a time.

Being realistic about the work that remains

This is worth setting out clearly, because it changes what leading one of these areas involves. Most of these features already work today. The gap is usually not the core capability. The gap is that we have not yet shipped the recipes and demo configurations that make them work end to end, with results we are happy to show, on one realistic site.

This is the classic "it works" problem. Yes, it works in isolation. Making it work completely, convincingly, and repeatedly in a demo is a different kind of effort. So the leadership across these areas comes in two shapes. Some of it is making it work in the demo, building the recipe, wiring it to real content, and tuning prompts and results until the output is good enough to present. Some of it is net-new implementation, where the capability still needs to be built or substantially extended. Both are genuine leadership, and we are clear with anyone stepping in about which kind of work their area involves.

The team that gets us there

None of this happens without the hard work of our contributors. To strengthen our governance model, we have chosen to place responsibility for each area in the hands of a dedicated lead, someone who owns its direction and keeps a clear, public backlog for contributors to work from. The sprints themselves do not change. What changes is that every area now has a clear owner, and together these leads form the Inside AI leadership team.

Workflow Drupal AI Leads

Here is where each Inside AI area stands today.

Two things are worth making explicit, because they shape whether you picture yourself in that table.

First, a name beside an area does not mean you cannot join this focus area. The opposite is true. We are looking for strong teams rather than single owners, so contributing to an area that already has a lead is as real an opportunity as taking on one that is open.

Second, there is no formal application process. Becoming a lead happens through the work itself. A lead prepares issues and keeps the backlog in good shape. Our delivery managers, Arian Raeesi and Vidit Anjaria, plan those issues into the sprints. Some go through joint grooming with the technical leads, Marcus Johansson and Artem Dmitriiev, to check they fit the architecture, and I stay involved as product lead to keep them aligned with the overall product direction.

If one of these areas interests you, reach out in the #ai-initiative channel on Drupal Slack, or to any of us directly, and begin. Taking on that responsibility is what makes you a lead, and it is how you join the Inside AI leadership team.

Come and build it with us

This is where AI inside Drupal stands today. We know what our partners want, because we asked. We know where we want to show it working, at Rotterdam, on one site. And we know the only way to get there is together, with a team that owns each part of it.

If you have read this far, there is a good chance one of these areas already appeals to you. Come and say so in the #ai-initiative channel on Drupal Slack, or explore becoming a Drupal AI Partner. The demo will be better with your help, and so will Drupal.