Written by
Artificial Intelligence is redefining how websites and digital experiences are built, managed, and experienced. This transformation shifts the focus from manual workflows to results-oriented AI orchestration, where users define goals and AI contributes to achieving them. In this dynamic context, Drupal positions itself as a proactive leader in AI integration, distinguished by its open-source and responsible approach and its robust community.
AI adoption in Drupal has seen remarkable acceleration. During DrupalCon Atlanta 2025, Dries Buytaert, founder of Drupal, highlighted how AI framework adoption has tripled in just one year, involving even large organizations like the European Commission and the United Nations, actively engaged in experimenting with Drupal and AI. Furthermore, in the AI Initiative announcement, over 290+ AI modules are already highlighted as available.
This trend is not about replacing human capabilities, but rather enhancing them, freeing up time for creative and strategic work while AI handles routine activities, suggests improvements, and responds to feedback in real-time.
After outlining the main innovations of the ecosystem in our article “The future of Drupal CMS 2.0: All Innovations expected in 2025", this deep dive focuses on the latest innovations in the AI field for Drupal, exploring in detail the functionalities of the Drupal AI module and the progress of the strategic Drupal AI Initiative. We will also outline SparkFabrik's active role in this evolution, demonstrating constant commitment and a clear vision for the future of the ecosystem.
The Drupal AI Module: The Heart of Innovation
At the center of Artificial Intelligence integration in Drupal lies the Drupal AI module, a unified solution that enables the use of various AI technologies within the platform. This module provides a complete framework for easily integrating AI into any Drupal site, supporting a wide range of models and providers. Its primary goal is to offer a suite of modules and an API that serve as the foundation for generating textual content, images, audio, video, translations, but also entire components and sections and much more.
The AI module's strength lies in its abstraction layer, which enables seamless integrations with third-party AI providers like OpenAI (ChatGPT, DALL-E), Anthropic (Claude and also Claude Code), Google (Gemini, Vertex), Perplexity, Fireworks, and Mistral, or more specialized ones such as DeepL and ElevenLabs. It's also possible to use open-source models hosted on user-controlled servers through integrations with Olama, LMStudio, and Huggingface, among others. This functionality is particularly appreciated by organizations operating in regulated sectors.
This flexibility ensures that the AI module is a powerful tool both for site-builders, who can create complex applications without writing code (we discussed Drupal's first-class low-code/no-code capabilities in a previous article), and for developers and administrators, who benefit from simplified integration and immediate support for multiple providers.
The Drupal AI module aims to be the foundation of AI functionalities, offering a set of standard tools to use directly or on which to develop custom integrations. Indeed, the core functionalities are already extensible with a series of sub-modules, recipes, and other compatible modules. To mention some, among these stand out:
- AI Core: Provides access to common AI models (ChatGPT, Claude, Gemini...) and is extensible to any needed model.
- AI Automators: These are various sub-modules that allow populating and modifying any field in Drupal. Each Automator has specific functionalities, such as automation, web scraping, OCR extraction, chart data extraction, summary generation, transcript production, email address extraction, etc. The great versatility allows creating complex AI applications, where different prompts and different automators can be chained in custom workflows.
- AI Assistants API + Chatbot: A framework for configuring chatbot functionality, enabling advanced forms of AI search and enabling conversational search.
- AI Content: Offers assistance tools for content creation and editing, such as precise tone-of-voice adjustment, summary creation, taxonomy suggestions, moderation violation control. This level of flexibility and control is fundamental for quality output that respects your brand guidelines.
- AI Translate: Provides AI translations with one click, ideal for multilingual sites. For translations, there are several other alternative modules, specific to your needs and configurations.
- AI Search: Improves traditional search with semantic understanding, allowing the search engine to understand the meaning of users' terms. To achieve maximum results and reduce hallucinations, it's necessary to use vector databases and provide access to your own data (RAG). SparkFabrik also proposes a semantic search solution based on Typesense.
- Vector Databases: Enhance semantic search, chatbot, assistant, and other features. They allow expanding LLM knowledge by extending it to proprietary business data, which is vectorized and saved in the database.
LLMs can then recall this data and generate more relevant responses through the so-called Retrieval Augmented Generation (RAG) process. Various vector databases are supported, including Postgres, Milvus, Pinecone, Azure, and SQLite. - AI Agents: A framework for creating custom AI agents (currently, text-to-action type) capable of manipulating the website based on provided instructions, both on the content side (e.g., text and image generation) and on the configuration side (e.g., creating taxonomies or new page types).
- ai_image_alt_text: Automatically generates alternative text for images, increasing accessibility and improving SEO. Relevant and descriptive alt-text is also a fundamental accessibility requirement to ensure compliance with the European Accessibility Act.
- ai_seo: Provides SEO analysis and reports for individual nodes. Thanks to targeted feedback, administrators, content managers, and marketing teams can optimize both technical SEO and semantic SEO of their content.
Thanks to AI-powered analysis, not only basic aspects are considered, such as structure, headings, meta tags, image optimization, and URLs, but also more advanced aspects like Topic Authority, Topic Depth, keyword usage and natural language, responsiveness, and loading times. In short, a 360° analysis that also includes new SEO aspects that emerged in the AI era. - ai_text: allows easily creating an ai.txt file to give instructions to AI systems that interact with the site (in a way completely similar to robots.txt for web crawlers).
- AI Providers: To enable AI functionalities, you obviously need to choose one or more providers and enable the related module, for example Anthropic, OpenAI Perplexity, EvelenLabs, DeepL Translate. Many providers are supported, from the most well-known to local models, ensuring maximum flexibility.
These integrations (only some among all those already available or under development) demonstrate the depth and breadth of AI capabilities that Drupal can offer, covering a vast spectrum of digital needs.
Recent Evolutions: Drupal AI 1.1.0 and 1.2.0
The Drupal AI module is constantly evolving, with releases that introduce increasingly sophisticated and user-friendly functionalities, add AI providers, and resolve issues. In the spirit of sharing typical of the Drupal community, on the occasion of major releases, the Drupal AI Initiative team details the latest news.
In particular, after the initial launch, we have already enthusiastically witnessed two main releases of Drupal AI: 1.1.0 in June and 1.2.0 in July. Recent innovations aim to further simplify interaction with AI for editors and developers, enhancing content creation, automation, and site management. Let's see the main ones in detail.
Content Power-up, Between Creation and Content Management
The new functionalities make content creation and editing more intuitive and AI-assisted:
- Field Widget Actions: It's now possible to call an AI function through special buttons that can be added to any field in the interface (entity form).
These allow editors to easily interact with AI: in one click they can generate content to populate fields and, if not satisfied, generate again or intervene manually, always maintaining control over the final output. This functionality integrates with AI Content Suggestions, AI Automators, and AI Agents modules, supporting a wide range of uses.
For example, it's possible to rewrite an attachment's filename so that it's relevant to the content (finally goodbye to dozens of "image01.png" in your system) or generate alt-text for an image. Furthermore, you can request assistance in generating titles and summaries, based on page content and adopting your personal style thanks to a custom prompt. It's also possible to select or generate system tags to classify your content, or extract specific information like email addresses.
Important, for each of these widgets it's possible to provide specific instructions and context, enabling the generation of truly personalized content.
To ensure flexibility for the most diverse use cases, multiple widgets can also be connected to each field (for example, imagine having two buttons to generate a 50-character summary or a 200-character one, or to generate images with different styles).
In this video, a complete overview of Field Widgets is available, including their configuration. - Content Suggestions extended to other entities: Content suggestion functionalities, previously available only on Nodes, have been extended to Block and Taxonomy Terms entities as well.
Additionally, AI content suggestions can be generated based on an entity's rendered HTML, therefore on its final appearance, role, and content (for example, for a button, a CTA will be suggested, not a multi-line paragraph). - New types of Automators: AI Automators now include new available types. Among these, "Image Alt Text" stands out (which analyzes an image and context to generate appropriate alt text), "Image Filename Rewrite" (to rename image files for SEO purposes), and "Summary Generation for Text with Summary" (to generate summaries from main text).
Thanks to the Field Widgets Actions we saw earlier, Automators can also be easily activated by editors, directly from the user interface.
Agentic Framework and Developer Tools
The keyword of AI's latest frontiers is "Agents". The introduction of the new agentic framework in Drupal and advanced tools for developers accelerates innovation and customization:
- New Agentic Framework: The big news of v1.1.0 is the introduction of the Agentic Framework, which allows anyone to build AI agents without writing a single line of code.
The AI Agents module simplifies and optimizes the creation of every type of AI agent, including "text-to-action" agents that can create or modify content types, fields, and taxonomies based on natural language instructions. This transforms users' simple words into concrete actions within Drupal.
A notable aspect is that agents are stored as configurations, making them exportable and reusable across different systems. They can be activated from various points, including Chatbots, CLI, widgets, and through APIs, making Drupal an ideal platform for agent execution thanks to its flexibility and stability.
Furthermore, Drupal's native support for the Model Context Protocol (MCP) allows agents to connect to potentially any tool, exponentially extending their capabilities. Equally importantly, human intervention will always be possible (the so-called "human in the loop"), ensuring human governance. - Prompt Library: Like agents, all prompts can now also be distributed as configurations, facilitating the provision of suggested prompts from third-party modules.
This is crucial for a well-functioning AI ecosystem: prompt engineering is a complex field, and a predefined library with best practices streamlines processes, enables users, reduces uncertainties and dependencies on maintainers. - Mocking Library for testing and replay of AI requests: This is a new functionality designed specifically for developers to facilitate development. Instead of making a real request, it's now possible to simulate a request to an AI provider and reproduce it during development or automated testing. This allows alleviating both long waiting times and high costs associated with prompt processing by AI providers.
- Support for more file types for AI models: The AI Core module base has been modified to support any file type (including PDFs and videos) to provide to AI models, in response to the evolution of LLM models.
- Recipes, abstraction from Providers and Vector Databases: Recipes are predefined, reusable configurations that simplify the implementation of specific features, such as automating module installation or configuration implementation (we go into more detail in this article).
Previously, Recipes that added AI features such as “chat with your documents” or similar required you to specify exactly which providers and vector databases were used, as well as certain settings. With the latest update, however, AI providers and vector databases have been “abstracted,” meaning they no longer need to be specifically defined at the Recipe level, as well as some common features. It's a small change, but it has made AI Recipes much more modular, versatile, and maintainable, as well as more reusable. - AI Agents Testing Tool: A new visual tool for testing AI agents has been added, allowing the configuration of complex scenarios and repeatedly retesting them without needing to be a developer.
Interface Creation: AI + Experience Builder
Experience Builder is the evolution of all previous Drupal site building solutions. Modern features like completely visual and drag-and-drop interface, integrated previews, and on-the-fly creation of new components make it a truly advanced solution for creating new digital experiences (we already talked about it here and here).
The integration of AI with Experience Builder has the potential to even more profoundly revolutionize not only the authoring experience, but also the design and development processes themselves in Drupal.
During the March 2025 Driesnote, it was demonstrated how an image from Figma can be transformed into a UI component in real-time. The user can interact with the AI agent through chat, requesting iterative changes and refinements to colors, button shapes, and other elements, with changes immediately visualized in a code editor integrated into Experience Builder.
Creating a single component is just the first step: the vision is to create entire pages or sites, ideally even with migration from other systems. In this vision, AI will help users build and perfect sites faster, more intuitively and creatively, while maintaining human oversight.
Enhanced Semantic Search
Search functionality in Drupal has been profoundly improved thanks to AI. The AI Search sub-module elevates traditional keyword-based search with semantic understanding, allowing the site to grasp users' intent rather than limiting itself to typed words. This translates into more relevant results, even when search terms don't directly match the content.
The overall impact can be truly significant, both in terms of user experience and economic return for the organization (think for example of product search in ecommerce, or information search on a university portal by a potential student).
This functionality uses two search indexes, one traditional and one semantic, powered by vector databases like Milvus or Pinecone, fundamental for reducing AI "hallucinations," those situations where AI produces inaccurate information but presents it with confidence. These vector databases contain data and information specific to your organization, chunked and vectorized. LLMs can then use this data to provide more precise and contextualized answers (or search results).
An example of technology that integrates with Drupal to enable these advanced functionalities is Typesense, an open-source search engine that offers AI-powered semantic search capabilities, providing faster, more accurate, and personalized results. The Typesense module is developed and maintained by the SparkFabrik team. You can explore this solution in the following presentations.
Overall, these innovations demonstrate a continuous commitment to improving usability, efficiency, and versatility of AI within Drupal, facilitating developers, making it accessible to a broader audience, and supporting increasingly complex use scenarios.
The Drupal AI Initiative: A Coordinated Drive for the Future
The Drupal AI Initiative represents an important strategic step for Drupal's future, channeling community energy in a coordinated and funded direction.
Although the ecosystem of Artificial Intelligence functionalities, integrations, and modules in Drupal is experiencing significant growth, as we have seen, to ensure bringing real strategic impact to the market, the need to go beyond simple voluntary contributions from contributors has been recognized. To build a truly powerful and competitive AI ecosystem, a professional and dedicated team is needed, focused only on Drupal AI.
From this awareness was born the Drupal AI Initiative, officially launched on June 9, 2025, whose goal is to bring structure, strategy, and a shared direction to AI innovation in Drupal. By ensuring continuity and quality, it guarantees that Drupal doesn't fall behind with the rapid evolution of AI in the market.
The central strategy is to fund a team of full-time contributors to accelerate AI innovation in the Drupal ecosystem. Over $100,000 has already been allocated to be entirely dedicated to the initiative, but the aim is also to attract sponsors, defined as "AI Makers." These sponsors will be able to contribute not only financially, but also by dedicating full-time human resources (at least ½ FTE, Full Time Employee), with the expectation that involved companies are fully dedicated to the coordinated development of the ecosystem.
The initiative leverages a clear organizational structure, with a Leadership Team dedicated to guiding product direction, fundraising, and collaboration between different work areas, fundamental aspects for ensuring coordinated effort at a global level. A funded Delivery Team, equivalent to several full-time roles, is dedicated to execution, including technical leads, UX, and project managers. Active work tracks cover key sectors like AI Core, AI Products, AI Marketing, and AI UX.
Initial progress has seen the initiative gain momentum, primarily with the release of the latest major releases of the Drupal AI module. On the marketing front, dedicated pages have been created on Drupal.org and collaborations have been initiated for a series of webinars with the European Commission. The team also participated in the AI Summit London, an important artificial intelligence conference, to present Drupal AI to a broader audience.
One goal is to organize regular webinars to keep the community informed about progress and raise awareness among potential clients and end users about Drupal AI's potential. The community is also constantly aligned through asynchronous meetings held every Monday in the #ai-contribute Slack channel, where all contributors can participate and align through messages for 24 hours, with conversations subsequently published publicly on Drupal.org.
Importantly, the initiative fully embraces Drupal's philosophy regarding AI, built on fundamental principles that guide innovation. The heart of everything is the conception of AI itself, whose role is to enhance human capabilities, not replace them. Based on this, the aim is to implement a framework that promotes responsible AI management, ensuring oversight and approval of workflows (Human in the Loop), audit trails, and compliance tools.
Furthermore, in pure Drupal spirit, innovation must be guided by the community's real needs, not by roadmaps and a priori decisions. For example, ensuring users full freedom of choice in terms of AI solution providers is considered fundamental, without providing constraints to specific vendors.
Furthermore, a recent survey invited end users to share their needs, contributing to directing development resources to functionalities felt by real users. At the same time, the community and agencies are the foundation of Drupal's success: developing a platform that grows in synergy with both end users and the agencies that support it therefore proves essential.
SparkFabrik's Role in the Drupal AI Ecosystem
SparkFabrik is not just an observer of AI innovations in Drupal, but an active and strategic participant in their development. With over a decade of significant contributions to Drupal's open-source ecosystem, SparkFabrik has always positioned itself at the forefront of its evolution.
It's clear how the advent of artificial intelligence is requiring a radical transformation of the entire development process, which now becomes increasingly outcome-oriented rather than time-dedicated, but is also opening completely new perspectives and evolutions throughout the ecosystem, with unprecedented speed.
This is why we have initiated an internal strategic initiative dedicated precisely to the Drupal and AI combination, dedicating ½ FTE for all of 2025 (50% of our expert Drupal developer Luca Lusso's time) specifically to projects in this area. This concrete commitment to strengthening and expanding the ecosystem fits perfectly with the global AI Maker Initiative and translates into constant and deep collaboration with the community.
In our first contribution sprint, we worked on two aspects we consider fundamental for the Drupal AI ecosystem: proposing an effective local development environment based on DDEV and beginning reasoning on how to integrate a "guardrail" system directly within LLM integration flows.
More in detail, regarding the local development environment, we noticed that several contrib modules adopt a pattern based on a DDEV add-on that replicates the functioning of GitLab's CI pipeline on Drupal.org. For many modules contributed by SparkFabrik, we have also chosen the same approach (Monolog, WebProfiler, ...). The broader community has not yet decided if this is the best path to follow, and for the Drupal AI ecosystem this way of working could lead to more complications than benefits.
Indeed, the need to add some DDEV-specific configuration files within the project repository and the way Drupal installation occurs locally make this approach less ideal in a context like Drupal AI, consisting of dozens of modules that potentially need to be tested and developed in parallel.
The solution proposed by Luca Lusso, and accepted by the community, instead saw the proposal of a new generic DDEV add-on, which can be used by any contrib module and capable of simplifying contributions to multiple modules simultaneously. SparkFabrik's solution resolves previous limitations, streamlining development and testing of all future contributions.
In parallel, we initiated reflection on integrating "guardrail" systems directly into flows involving Large Language Models (LLM). This aspect is crucial for ensuring responsible and safe use of AI within Drupal, ensuring that model output is always aligned with preset values and objectives.
Another tangible example of SparkFabrik's commitment is active participation in meetings asynchronous in the #ai-contribute channel on Slack. These meetings, which take place every Monday and last 24 hours, allow global alignment through messages, and all conversations are subsequently published openly on Drupal.org.
Luca Lusso's constant presence in these discussions (for example, see conversations from June 30, 2025, July 7, 2025, and July 14, 2025) provides SparkFabrik with direct access to cutting-edge information and strategic discussions, allowing us to anticipate future developments and influence Drupal AI's roadmap.
In this scenario of rapid evolution, SparkFabrik's commitment to developing the Drupal ecosystem, research on AI and its practical application, is not just a technological investment. It rather represents a clear affirmation of our position as a strategic partner and thought leader, well beyond the role of a simple service provider.
Conclusion: Drupal AI, an Intelligent, Open and Collaborative Future
Drupal is actively leading the Artificial Intelligence revolution in the world of Content Management Systems, adopting an approach that is inherently open-source, human-centered, and highly flexible.
Starting from an already excellent foundation, the Drupal AI Initiative is significantly accelerating the development of powerful and responsible AI tools, thanks to a clear strategic vision, dedicated funding, and a rapidly growing community.
From the unified AI module, with its vast range of sub-modules and supported providers, to innovations introduced in recent versions, Drupal offers cutting-edge functionalities that transform content creation, site management, and optimization of digital experiences.
These developments not only improve efficiency and quality of web projects, but also reinforce Drupal's fundamental principle of maintaining people's centrality and human governance, ensuring that AI is a tool for enhancement and not replacement.
SparkFabrik is proud to be an integral part of this transformation, not limiting ourselves to adopting new technologies, but actively and strategically contributing to their development and strengthening our position as thought leaders in the Drupal ecosystem.
Our dedicated internal initiative, significant resource commitment, participation in the community and strategic discussions, our expert Drupal developers' commitment to directly contributing to code demonstrate our vision and ensure that SparkFabrik is not only at the forefront, but contributes to defining the future direction of AI in Drupal.
If you wish to explore AI's potential for your Drupal site, or if you are considering implementing a Drupal environment, SparkFabrik is available to support you in navigating and excelling in this new and dynamic digital landscape.
We invite you to:
- Explore SparkFabrik’s Drupal services
- Consult our case studies that illustrate implementations in various sectors
- Contact our team for an evaluation of your specific context and objectives
Explore more features and aspects of the Drupal platform in our dedicated articles.
