Drupalcon Atlanta 2025

Professional Article
April 6, 2025

Hey there, this is my Drupalcon-experience review, all the views/opinions here are mine and don’t extend to the company or partners I work with.

This was my second Drupalcon in general, and first one in North America - I used to be cautious about attending Drupalcons because the pricing can be quite prohibitive for small businesses, freelancers, and independent devs… I do enjoy DrupalCamps though - my 1st being BADCamp (US), then followed by camps in Ukraine, Romania and the ones we organized in Moldova, I think I attended ~6 camps and co-organized 3 more.

This year, thanks to Dropsolid, I had the chance to visit the US-flavored Drupalcon. This time it was in Atlanta, Georgia - in the Hyatt Regency Hotel. The venue was pretty fancy, I’d say - classy. I liked it, although not sure it was called for, but okay, it was a pleasant stay.

Overall the camp lasted ~4 days:

  • 1st day preparations + opening,
  • 2nd day the actual day of the conference,
  • 3rd day was another day full of sessions and what not, and
  • 4th day was just the code-sprint (contribution day).

There were many companies present - some are well known in the community (most) and few were less known.

I don’t have the precise number of attendees, but I think it was ~1000 people more or less (well, ChatGPT estimated ~700 people in this photo 👇). For some reason I expected it would be something like 1.5k - 2.0k (a friend told me it used to be ~3000 attendees!).

(photo from: https://herchel.com/articles/drupalcon-atlanta-pics-tidbits-takeaways - idk where to find other group pics - thanks mate!)

I also expected to see some non-Drupal attendees: either students, or B2B companies, or governmental organizations. It was 95% hardcore Drupal companies (and their employees) visiting other Drupal companies’ booths. I think I only met 2-3 people that weren’t hardcore Drupal-involved. Even product-focused folks from Automattic (yes, the WP company) were Drupal-involved architects.

If you don’t want to read the rest, the TL;DR is - I liked the event, but I can see why others won’t.

I liked it because I’m ~14-15 years working with Drupal, many people that I met online, or in interviews, or in emails or slack - were there in-person: Amitai (Gizra), Ryan (ex-Commerce Guys), Marcus (the one building 90% of all the AI-module ecosystem), Alexei (my ex-colleague, now at Jakala), our dev-friends from Peru, etc - it’s fun to interact, debate, challenge ideas, approaches, etc. At times, the atmosphere felt somewhat corporate and exclusive (closed), lacking the broader, open-community vibe that drew me into Drupal in the first place - but I still felt good as it's "my waters".

I’ll try to go over the Announcements (that I paid attention to), Positives (good vibes), Negatives (things we can enhance) and Moving On (or next steps).

Announcements

Look, I’m sure there were a ton of things announced, but I’ll keep things short-ish.

  1. Progress in XB (Experience Builder) - a LOT of people wait in excitement to see how this initiative progresses. I personally really liked that we embraced Tailwind CSS and ReactJS for components. I dislike when we try to reinvent the wheel in our Drupal bubble and I like when we embraced jQuery back in the days (ah, this will be mostly the theme of this article 😀)

  2. Site Templates - an unknown (really, it’s not defined yet) wrapper around recipes. Will it be a better version of Distributions? Will it be something totally amazing? Most of these are unknown. Although there was 1 “site-template” already created during the conference by Ryan (the one from Drupal Commerce) - it was still just a recipe.

  3. Potential Marketplace - there were talks (Birds of a Feather sessions, or BoFs) around this idea of creating a unified / central Marketplace for these site templates (which are NOT defined yet).

  4. AI module(s) - the new 1.1 branch will finally use function calling - but this bit is yet to be polished, tested and released (it’s under development - a moving target now).

I really liked the progress in XB - I loved that they (although in an opinionated way) embraced other stuff that’s been popular out there (and for years).

I like the idea of Site Templates and (a bit less) of centralized Marketplace. AI module ecosystem - I have mixed feelings. I like some of the progress, but not the overall trend. I’ll explain these in detail in Positives vs Negatives - down below.

The Site templates with unknowns + the marketplace.

Positives

Like I mentioned above - I really liked to meet people I knew for years (and lost touch with) and people I knew or talked to online, but never met. If the barrier to join would be lower, I’m sure I’d see even more of such people, but I was happy to see the ones I did too.

AI-landscape - a lot of companies embrace AI. Some offered AI-powered Translation Services (as a product), others offered a custom “Provider” for the AI.module (Amazee) (which will probably be a private instance of Claude), other companies just offered services in building simple RAGs and/or Chatbots. There were rumors of some companies using Claude Code and getting a very good yield out of it.

There was even a demo of someone using AI to build out something in Experience Builder - but QED42 folks published this video over 1 month ago:

After talking to companies and developers - I was pleasantly surprised that there are more people that share my thoughts on AI, and how it can be integrated within Drupal websites - especially when it comes to workflows and orchestration. That was proof to me - that the community is big and that there’s room for multiple approaches and ideas, as it was always the case - in example: RESTful core vs RESTful (Gizra) vs JSON:API vs GraphQL, etc.

Negatives

Organization-wise - it’s too expensive and I think the community is losing because of it. Some will point out - “go to Drupal Dev Days or to your local Camps, those are more affordable” - sure, but that is not ‘the answer’. If I’d be a small 5-10 people Drupal shop that wants to look for some clients or devs or mingle around - I can’t afford to spend thousands of dollars on a ‘con - also even if I can - all I’d see are 800-1000 of other folks like me, walking around and talking to ourselves.

There are huge risks with this: events become less attractive for business and for developers. Our Drupal-bubble becomes tighter and smaller, and eventually smaller companies and developers find themselves looking for alternatives. If there’s more friction and less mingling (i.e. come for the code, stay for the [increasingly-expensive-to-be-a-part-of] community) - it’ll be easier for people to leave. How to tackle this - see Moving On section 👇.

I noticed very few, if any, attendees under 30 (I’m closer to my 40s and I felt young there). This might reflect the cost barrier and the overall difficulty attracting newcomers to Drupal. It’s something we could address by offering more youth-focused pricing and outreach (anyone remembers Drupal Global Training Days? 🖐️ What happened to them? We used to sneak into Universities and glue DGTD posters/flyers and attract curious people into our community...).

And then the AI landscape - there were some presentations that mentioned Drupal as a leader in CMS and AI, although the examples showcased - like a basic RAG, felt a bit outdated for 2025 standards, sure, RAGs were a thing 1-2 years ago, but today it’s definitely not an example of being a leader / edge-case.

We had our 1st RAG up and running, in June, 2023 (that was 4 days before function calling was announced by OpenAI) - and it worked on Langchain with Drupal 7. Around 6 hours after the announcement there was a Langchain update that already had function calling agents and examples - and I rewrote our agent to use that.

Or, another example, in August, 2023 - we had a crawler that dynamically crawled websites, built a vector-knowledge database from each of these, used a special RAG approach to query that knowledge-base and build-up full medical profiles of clinics: medical services, descriptions, address, etc. You’d understand why I am saddened to see that RAG is the use-case for Drupal being a leader when it comes to AI and CMS.

Back to Drupalcon ATL - assuming we’ll build a perfect AI-to-XB integration - my concern is that we may end up reinventing the wheel, creating a less robust version of existing solutions like Vercel’s v0 (or many similar tools that are out there: Bolt, Webcrumbs, Tempo, TailwindAI, etc...). I’m not convinced this approach will appeal to CxOs or marketing professionals who typically prefer well-established tools for their upcoming campaign. Nor do I think developers servicing those people will want to use in-browser code-editor (and not their VSCode or PHPStorm), and small floating text-boxes with little to no control - to help them create those elements. Most will stick to VSCode, Cursor, PHPstorm - and will use AI within these tools to craft those (or use CLI tools like ClaudeCode).

Even when it comes to filling the fields with AI generated data - most people that write articles, they don’t want AI slop, they want to get real stuff, and they’d rather stay within tools that can iterate over and over until they are really happy with their content. So, 1 click widget to generate XYZ text-field content - might be good for demo purposes, but almost never is good enough for real production use-case.

Moving on

I’ll give few ideas on various fronts:

Organization-wise - we should take it a notch lower, I’d be fine to be in 2-or-3 star-hotel or even an Airbnb, and have the event hosted in a University Campus or some Government-sponsored event venue - but to have tickets half-the-price; have at least 300-500 student tickets for symbolic price of 50$ or so, have a special business/gov entrance price at a lower tier (200-300$?) so that more business / gov. agencies can come into the venue.

Make partnerships with companies to bring the price of organizing the event low 👇 - there could be an internet sponsor, a coffee-break sponsor ☕, a party-sponsor, a-print-company-sponsor, etc. You could help promote national/local businesses and get stuff for a huge discount or even free.

I noticed there’s a third-party organizer for these events. While this approach has advantages - such as specialized expertise and logistics - we may also miss out on building a lasting knowledge-base over multiple Drupalcons and strengthening direct partnerships with local vendors. For example - we lose maintaining coffee-partners ☕ or print-partners and how to work with them, how to find better deals to bring the cost of the whole thing down.

AI-wise - I’ve mentioned before, I really like when Drupal stays in its “lane” - we don’t invent a new CSS framework, we embrace Tailwind; we don’t invent a new JS framework, we embrace ReactJS (and previously jQuery). But when it comes to AI - even though there are hundreds of AI orchestration frameworks, with thousands of active, product-focused and hungry developers - here, we decided to rewrite everything in PHP and make our own flavor. It’s still a valid approach, but I would really want to see Drupal embrace other third parties - delegate the workflow and orchestration to a library (or more) and act as a really good MCP-server (so that other agents can do Drupaly-stuff within Drupal).

(ChatGPT's version of us navigating the map to success, avoiding the Valley of no Projects and the No-Devs Deserts of Cobol)

Drupal’s strong points are its entities, fields, form-api, render-arrays, low-level APIs and integrations - it’s meant to build strong CMS platforms. This is what other modern frameworks “steal” from Drupal - a signal that these are our strong-points.

Drupal community looks up to its old foes: Wordpress, Joomla, Magenta. But there are new kids on the block: Strapi, Payload CMS, Contentful, Ghost, Sanity, DatoCMS (yes, most are headless, but that’s because there are really unbeatable good front-end frameworks). I think we should take a step and look outside of the bubble.
Sidenote: I also noticed some folks feel anxious about having to learn React, TypeScript, or Python. They wonder if they can simply stay within a familiar PHP environment for AI tasks. While that’s certainly possible, it’s worth acknowledging the exciting developments happening in other languages and ecosystems (read: outside of our bubble).

We sometimes position ourselves like a traditional premium brand - think BMW - competing alongside innovators like Tesla or BYD. On some fronts, we’re a bit behind these new players - these new players learned everything they could from us, took the best and ran with it (things like entities, fields, revisions, moderation, etc) - and added a lot of new things.

I’m not saying anything like Maxime Topolov did - I think we should just embrace our strong points and delegate the weaker ones to other frameworks - embrace those and commit to those fully. And it’s not even us having weak points, it’s about having way stronger alternatives out there - especially those which already implicate thousands or tens of thousands of developers.

Someone asked me - “Where would you want to see Drupal in the next 6 months or so?” - Here’s what I think:

  • Having XB out there - is very important. Also to be aware how it will be used by actual developers - i.e. very unlikely devs will go into the browser to write JSX and other stuff - we like our plugins - be it highlighting, AI, or other things. So, I wish to see some well-thought progress in the XB domain.

  • Figure out Site Templates - this is a very tricky one - it’s a very powerful tool, but also very tricky to get it right. If we accept that some.php can be proprietary, many will decide to move their forms and logic there, just to charge $, and this will have ripple effects in our community - similar to WP. SQL queries should NOT live in the template, but if the template can have proprietary files - they will move there 100% 🙂 I’m curious to see where we will get in 6 months - but I hope a decision will be made.

  • AI - I don’t think we have to place all our eggs in this basket “Drupal is the AI workflow management system”. We should acknowledge that real production systems - will most likely want to use other AI frameworks - and we need to move Drupal forward in a way that would allow it to integrate with any of those, and not compete with them. In 6 months - I’d like to be able to pick a system off the shelf - i.e. LlamaIndex or Langchain or Vercel’s AI SDK - and be able to easily integrate it with a Drupal 11 website. This probably translates to - a 💪 strong MCP Server (with proper authentication) and some smaller unexisting (yet) modules - that can also play along with existing ai.module(s) effort but also offer alternative development paths.

Overall, I’m hopeful that these changes - both organizational and technical - can keep Drupal modern and accessible.

Btw, I’ll try to prep a session for next Drupalcon with examples on other AI paths - i.e. how it can be used with Langchain (and MCP) - most probably I’ll do a (or a few) blog posts about it before. 🫡

Thanks again to Dropsolid for the opportunity to visit the event, mingle with awesome people, talk code until 2 am - showing various AI PoCs, what's happening outside of the Bubble, and more. Also, thanks to Alexei for hanging out and exploring the 'con - it was awesome to see you there 🤘

Let me know your thoughts - whether you agree, disagree, or have alternative ideas. I’m curious to hear other takes on Drupalcon’s future, especially around pricing, AI, and building a more open ecosystem. 

Comments:

Feel free to ask any question / or share any suggestion!