Drupal AI 1.2.0-alpha2 Released with Stability Fixes, New Features, and Token Usage Update

Drupal AI 1.2.0-alpha2 has released, introducing a wave of stability improvements and new capabilities that bring the project closer to a production-ready state. The announcement, shared by Marcus Johansson, outlines technical enhancements aimed at improving automation, accessibility, and developer flexibility across Drupal AI workflows.

The release includes fixes for bugs reported since the 1.2.0-alpha1 version, along with refinements to existing features. As expected for an alpha release, no upgrade paths are available, and further updates are planned ahead of the beta.

A standout feature in this version is the Views Automators Type, which allows AI Automators to access and execute Drupal Views from within workflow logic. This enables powerful new use cases, such as generating natural language summaries of top-performing content, creating personalized related content blurbs, or assembling editorial digests filtered by topic.

The Field Widget Actions module has also been enhanced. Editors can now use AI-powered buttons for a wider range of field types including select lists, numeric fields, and email inputs. Accessibility gaps in the earlier release have been addressed, making the interface more inclusive and responsive.

Support for streamed responses has been significantly improved. Developers can now use streaming while still benefiting from post-response actions like logging, token counting, and callbacks. This resolves a long-standing architectural limitation in PHP environments.

Improvements have also been made to the CKEditor plugin. The update refines how selected text is handled when invoking AI features inside the editor, providing a more intuitive and efficient experience for content creators working within rich text fields.

The release introduces normalized token usage for chat-based AI clients. Previously, token usage was exposed as raw data, which made it difficult for developers to implement accurate tracking or cost monitoring. With the new normalization system, third-party libraries can more easily manage usage costs and enforce limits. Work on a supporting module that leverages this system is already underway.

This update reflects the collaborative efforts of a global community. Dozens of contributors participated in the release, and support came from a broad group of organizations including 1xINTERNET, Acquia, Amazee Labs, Dropsolid, EPAM Systems, CivicActions, Drupal India Association, and institutions such as the European Commission.

For complete release details, documentation, and ways to contribute to the Drupal AI Initiative, visit the project page.