AI Agents can make your Drupal site feel smarter and more helpful. In this session, we'll build a practical customer assistant using Drupal's Tool API, MCP integrations, and no-code configuration, complete with automated Slack updates.
You should be comfortable with basic Drupal concepts like modules, plugins, and site configuration. No AI experience is required, just curiosity and a bit of developer spirit.
In this session, we'll take a hands-on look at how AI Agents can enhance real Drupal projects. We'll start by defining the goal of our demo: building a simple AI Customer Assistant that helps users find the right product and posts a Slack message whenever someone completes a purchase.
We'll walk through the toolset step by step. First, we'll see how to create a custom Tool using the Tool API https://www.drupal.org/project/tool, keeping the implementation clean, well-documented, and easy to maintain. Then we'll explore the MCP Client module https://www.drupal.org/project/mcp_client and how it brings external Tools from an MCP Server directly into Drupal. Along the way, we'll talk about why thoughtful Tool design, good documentation, and solid testing matter even more in AI-driven setups.
Finally, we’ll switch to a no-code flow and build the actual AI Agent using Drupal’s configuration UI. You’ll see how simple it is to wire everything together while still keeping guardrails and clarity at the center of the process.
By the end of the session, you'll be able to:
- Understand how Tools, Agents, and MCP integrations fit together inside Drupal
- Code a custom Tool using the Tool API with clean structure and clear documentation
- Use the MCP Client module to consume Tools exposed by an external MCP Server
- Build a no-code AI Customer Assistant that recommends products and triggers Slack notifications
- Appreciate how writing good instructions and documentations
