Paul C of WP Tuts walks through the AI features shipping with WordPress 7 and the companion AI Experiments plugin. The video separates the underlying infrastructure (centralized AI connectors for Anthropic, Google, and OpenAI) from the user-facing experiments (title generation, image generation, alt text, excerpts, review notes, and content summarization), then tests each feature live in the editor to evaluate real-world usefulness.
A centralized settings area in WordPress 7 where you enter API keys for Anthropic, Google, and OpenAI. Any plugin or tool that needs AI access reads from this single location rather than managing its own credentials. This is a significant architectural decision that reduces configuration sprawl and makes provider switching straightforward.
A separate plugin that adds user-facing AI features to the WordPress editor. Each experiment can be individually enabled or disabled. Current experiments include title generation, alt text generation, excerpt generation, image generation, review notes, and content summarization.
A tool in the WordPress admin (under Tools) that lists all available AI abilities, their slugs, and lets you test them. Useful for developers and site administrators who want to understand what AI capabilities are active on their installation.
Generates three alternative title suggestions based on your current title. You can select one, then regenerate again for further variations. Useful as a brainstorming aid, though Paul notes it would benefit from inline editing and more iterative control.
Reads your post content and generates an image automatically. Alt text is added to the generated image. The main limitation is lack of user control over the prompt -- it decides what to generate based on the content rather than letting you describe what you want.
Available when adding an image block, this option lets you write a custom prompt and generate an image directly in the editor. More useful than featured image generation because you control the prompt. Alt text is auto-generated for the result.
Reads your post content and produces a summary excerpt. Can be regenerated for different variations. Paul highlights this as particularly useful for writers who are good at long-form content but struggle with concise summaries.
Scans all content blocks and produces editorial suggestions (e.g., SEO improvements, vague headings). Suggestions appear as conversation threads you can reply to. The main gap is the absence of suggested replacement text -- it tells you what is wrong but not how to fix it.
Generates a summary block inserted at the top of the post. Very fast compared to other features. Useful for adding TL;DR sections to long-form content.
Paul emphasizes the distinction between features that look good in a demo and features that survive real workflows. His overall assessment is positive but measured: the AI experiments are useful starting points rather than finished tools. The centralized connector architecture is the most important long-term contribution because it creates the foundation for every future AI integration in the WordPress ecosystem.