The latest update to the WordPress AI Experiments plugin has finally landed, and it is a clear sign that Core is moving past the “chatbot gimmick” phase. While most of the ecosystem is busy wrapping fetch() calls to OpenAI in messy theme functions, the Core AI team is actually building architecture. Consequently, version 0.2.0 feels less like a demo and more like a foundation.
If you have been following my take on integrating the WordPress AI client, you know I value stability over hype. This release introduces meaningful editorial tools and, more importantly, the “Abilities Explorer” for us developers. Specifically, it addresses how we manage AI-powered actions without turning our functions.php into a prompt-injection nightmare.
AI-Powered Excerpts: Beyond Automation
The headline feature in WordPress AI Experiments 0.2.0 is the new excerpt generation tool. Many site owners treat excerpts as an afterthought, often leaving them blank or letting WordPress grab the first 55 words. However, this experiment allows editors to generate a neutral, concise summary directly within the sidebar.
From a dev perspective, the “naive approach” would be a simple button that replaces the text area. Instead, the team implemented a reviewable workflow. Authors can generate, edit, or reject the AI output. Therefore, the human remains the final filter—a pattern we should all be following in client work.
The Abilities Explorer: A Debugger for AI
For those of us building custom blocks or extending the editor, the “Abilities Explorer” is the real star. It is a new admin screen located under Tools > Abilities Explorer. This screen surfaces every registered AI Ability in the system. Furthermore, it allows you to test inputs and outputs directly in the browser.
In previous versions, debugging how a specific prompt was being handled by the official plugin was tedious. Now, you can verify if your custom ability is correctly registered and see exactly what the model returns. This is a massive win for transparency in WordPress AI Experiments.
<?php
/**
* Example: How to register a custom Ability in WordPress AI Experiments.
* Avoid hardcoding API calls; use the registered abilities framework.
*/
function bbioon_register_technical_summary_ability() {
if ( ! function_exists( 'wp_ai_register_ability' ) ) {
return;
}
wp_ai_register_ability( 'bbioon_tech_summarizer', [
'label' => __( 'Technical Summary', 'text-domain' ),
'description' => __( 'Converts complex code descriptions into readable summaries.', 'text-domain' ),
'prompt' => 'Explain the following code logic for a non-technical project manager:',
'capability' => 'edit_posts',
] );
}
add_action( 'init', 'bbioon_register_technical_summary_ability' );Architecture and Performance
Under the hood, the team is refactoring the framework to TypeScript and adopting modern @wordpress/build tooling. They are also moving toward centralized settings using DataViews. Specifically, this release lays the groundwork for upcoming experiments like Alt Text generation and content summarization for long-form posts.
Look, if this WordPress AI Experiments stuff is eating up your dev hours, let me handle it. I’ve been wrestling with WordPress since the 4.x days.
Standardizing the AI Interface
The takeaway here is simple: stop building custom, one-off AI integrations for every site. The WordPress AI Experiments plugin is defining the “WordPress way” to handle models. By using the Abilities API, you ensure your code stays compatible with whatever LLM the user chooses to connect—be it OpenAI, Claude, or a local instance of Llama via Ollama. Consequently, your themes and plugins remain portable and future-proof.
