John Tran Demonstrates AG-UI Approach for Agentic Drupal Interfaces

Agentic Drupal interfaces were the focus of a session by John Tran, CTO of ImageX, who demonstrated how AG-UI can connect AI agents with Drupal-rendered components instead of limiting responses to chatbot text.

The session, titled Beyond Chatbots: Creating Smarter, Personalized Experiences in Drupal with AG-UI, used a proof of concept developed with the University of Waterloo. Tran presented the work as an extension of semantic search, with Drupal remaining the source of truth while agents decide when to trigger structured interface elements.

Tran said the semantic search layer analyses intent, context, and meaning, then processes Drupal content through embeddings, vector storage, and retrieval-augmented generation. He described AG-UI as an agent-to-user interaction protocol that is bidirectional, event-driven, transport-agnostic, and agent-agnostic.

The demonstration showed a student advisor workflow using Drupal-indexed content, external weather and map calls, media carousels, programme cards, pricing tables, event listings, and Drupal forms. Tran said Single Directory Components used in Canvas or Layout Builder could also be rendered through the AG-UI workflow when the agent determines that a component fits the user’s request.

During the question period, Tran said ImageX was working on two pure Drupal semantic search implementations and was trying to reduce reliance on third-party agentic frameworks for that layer. The advisor work was presented as a proof of concept rather than a production release, and the session did not provide performance data, deployment instructions, or a release timeline.