Course · Advanced

AI in the Gutenberg Block Editor: Build AI-Powered Blocks, Block Bindings & Editor Extensions

Bring AI directly into the content-creation experience. Build a real AI-powered block, a custom Block Bindings source, an AI sidebar panel, a toolbar action, and an inserter variation, all running as JavaScript in wp-admin rather than as backend automation.

8 lessons
Start course →

Every course before this one in the track puts AI on the backend or agent-facing side of WordPress: abilities an MCP client calls, a PHP AI Client SDK call your own plugin code makes on the server, an external assistant reaching in over the REST API with an Application Password. This course is different. It’s the first one that lives entirely inside the block editor itself, JavaScript running in wp-admin, authenticated as the logged-in user through the ordinary cookie and nonce session the editor already has, no MCP server and no external client anywhere in the picture. The output is AI features an author actually sees and clicks while writing a post: a button that rewrites a paragraph, a sidebar panel suggesting tags, a toolbar action that summarizes the selected block.

This course builds directly on WordPress AI Foundations for the general mental model of the WordPress AI stack, and on LLM Integration in WordPress With the PHP AI Client SDK for the server-side half of every feature here, since each block or panel you build still needs a WordPress REST endpoint behind it that calls wp_ai_client_prompt(). If you’ve taken both, the PHP side of every lesson will feel familiar. That said, this course is written to stand on its own for a block or theme developer who already completed Course 1 and is comfortable with register_block_type(), JSX, and @wordpress/element: every server-side call is shown in full, not assumed.

By the end you’ll have shipped a working AI-powered custom block, a custom Block Bindings source, a document sidebar panel, a block toolbar action, and a block variation that puts an AI-preconfigured block into the inserter, five different, real patterns for putting AI in front of an author rather than behind the scenes.