The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →WordPress AI is not one feature switched on across every site. It is an emerging stack: WordPress core APIs, separately installed plugins, provider connectors and AI services. Today, that stack can help with tasks such as alt text, summaries, translation, moderation and image generation—but what a site can use depends on its WordPress version, editor, installed plugins, provider connection and permissions.
How is AI being added to WordPress?
Think of WordPress AI as a set of parts that can work together, rather than a single built-in assistant. Core provides shared building blocks; plugins add workflows; connectors link those workflows to providers; and AI services process requests. Installing one part does not automatically install or enable all the others.
| Layer | What it does | What it means for a site owner |
|---|---|---|
| AI Client | A WordPress core component for sending requests to AI providers and handling responses. Learn WordPress lists it as included in WordPress 7.0 and later. | It is infrastructure for plugins and features to use, not by itself a ready-made writing assistant. |
| Abilities API | A core registry of named capabilities that agents can discover and call. Learn WordPress lists it as included in WordPress 6.9 and later. | It gives software a way to describe and expose capabilities; it does not automatically give an outside agent access to a site. |
| AI plugin | A separately installed official plugin offering ready-made experimental features and a reference implementation. | You choose whether to install and use its features. Its current editor boundary is important: it supports the Block Editor, not the Classic Editor. |
| MCP Adapter | A separately installed bridge that can expose site abilities to external AI agents through the Model Context Protocol. | Access depends on the adapter, registered abilities, permissions and site configuration. It is not the same thing as WordPress.org’s plugin-directory MCP server. |
| Connector and provider | A connector links WordPress to a provider API; the provider performs the AI work. Local options such as Ollama and llama.cpp are also described in WordPress learning material. | Hosted providers may charge for usage. Local models avoid per-request provider charges but require capable hardware. |
These layers are related but distinct. For example, a site may have core APIs without the official AI plugin, or have a third-party plugin that connects directly to a provider. Check the requirements of the specific plugin and feature rather than assuming that a recent WordPress installation includes every AI workflow.
What can WordPress AI do today?
The official WordPress AI plugin presents a broad set of experimental editorial and site-management features. Its listed examples include:
#1 Best Overall
- Writing and editing: summaries, excerpt generation, title suggestions, meta-description suggestions, translation and editorial notes or updates.
- Images and media: alt-text generation, image generation and editing, and content resizing.
- Review and organization: comment moderation, content classification, and an Abilities Explorer for examining capabilities.
- Extending workflows: custom abilities, connector approvals and AI request logging.
The list is subject to change: the plugin is explicitly an experimentation lab and reference implementation, not a promise that every listed feature will remain unchanged or become part of WordPress core. Its design emphasizes practical experiments, user control and manual review before automation. WordPress describes the work as exploring capabilities that might merit consideration for core; that is not a commitment to include any particular feature.
What happens when a feature is used?
Provider-backed features need a route to an AI service. WordPress describes connectors as the bridge to provider APIs, with credentials stored centrally and selected features enabled by the user. A connector is not the same as the provider: it handles the WordPress-to-service connection, while the external service processes the request. Local model options change that arrangement, but require hardware able to run the model.
In practical terms, a generated suggestion is an input to an editorial decision, not proof that the result is accurate or suitable. Review text, translations, moderation decisions, image outputs and metadata before relying on them—especially when an error could affect customers, public information or site operations.
Rank #2
Is the official AI plugin safe to use on a production site?
WordPress.org labels the plugin experimental and advises testing it on a staging site first. Its FAQ says, “All AI features are opt-in and require manual triggering – nothing happens automatically without your approval.” That describes the plugin’s stated defaults; it does not establish that every provider, connector, configuration or output is risk-free.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Before enabling a workflow, identify which provider will receive requests, what content the feature needs, who can trigger it, and whether the result needs review. Use connector approvals and permissions deliberately, and test the workflow against your own content and setup in staging. The available official material establishes those opt-in and manual-approval mechanisms, but does not settle every privacy, security, reliability or output-quality question for every configuration.
Which AI providers are supported, and what does it cost?
WordPress learning material names connector options for Anthropic, Google and OpenAI. It also discusses local options such as Ollama and llama.cpp. Connector availability and requirements can change, so check the current connector and plugin listings before choosing an implementation.
The official AI plugin is free software, but that does not mean every AI request is free. Provider usage may incur separate charges under the provider’s terms; the cited WordPress material does not establish a universal price. A local model can avoid per-request charges from a hosted provider, but shifts the requirement to capable hardware. Confirm the provider’s current pricing and billing model directly before enabling usage.
Does the official AI plugin support the Classic Editor?
No. The official AI plugin currently focuses on the Block Editor and does not support the Classic Editor. Its maintainers say there are no plans at this time to add official Classic Editor support. Sites that rely on the Classic Editor, or on editing flows outside the Block Editor, should treat this as a compatibility limitation and look for a tool designed for their workflow rather than expecting the official plugin to add that support.
Recommended Free Tools
What else is in the WordPress AI plugin ecosystem?
The WordPress.org AI-tag directory shows that AI-related extensions reach beyond text generation. Categories visible there include chatbots, AI administration, SEO, automation, commerce operations, knowledge bases and provider connectors. Examples include AI writing and automation tools, agent-based site management, WooCommerce bulk editing, customer chat, and documentation or knowledge-base search. These listings are discovery examples, not endorsements or comparative tests.
The directory’s active-install counts are useful only as dated signals of adoption. On 2026-09-30, its listings showed:
| Directory listing | Active installations shown | How to read the figure |
|---|---|---|
| AI Engine | 90,000+ | WordPress.org directory snapshot accessed 2026-09-30; not a measure of quality or performance. |
| Official AI plugin | 50,000+ | WordPress.org directory snapshot accessed 2026-09-30; counts can change. |
| SiteGround AI Agent | 1+ million | WordPress.org directory snapshot accessed 2026-09-30; the count is for the listing, not a measure of AI outcomes. |
| All in One SEO | 2+ million | WordPress.org directory snapshot accessed 2026-09-30; its listing mentions AI SEO tools, but the count is not an AI-feature usage count. |
The directory is a discovery page, not an independently audited market survey. Installation counts do not show user satisfaction, output quality, active use of AI features or how a plugin performs on a particular site.
Example: AI Engine
The AI Engine listing describes chatbot and framework functions and says users need their own API keys. It also warns of possible conflicts with some caching or performance plugins. Treat that as a compatibility point to check against your own stack, not a general finding that the plugin is unsafe or unsuitable. Directory listings do not provide a controlled comparison between plugins.
How should you choose an AI plugin or workflow?
Start with the task, then check the integration and operating requirements. A tool for translation is not interchangeable with a chatbot, a commerce bulk-editing workflow or a developer tool.
- Match the job: decide whether you need writing or editing, image and alt-text work, translation, moderation, chat, SEO, automation, commerce operations or development support.
- Identify the layer: determine whether the feature comes from core, the official plugin, a third-party plugin, a connector or an external service. One product may combine several layers.
- Check editor compatibility: verify support for the Block Editor, Classic Editor and any other editing flow your team actually uses.
- Understand the provider model: establish whether you need an API key, whether usage can cost extra, or whether a local model and capable hardware are required.
- Review control and access: check whether features are opt-in, manually triggered, subject to connector approval or exposed as abilities to an external agent.
- Assess maturity and compatibility: confirm the current feature list, supported WordPress versions, maintenance status and known conflicts. Test consequential workflows on staging before production use.
How is AI changing plugin development?
AI is beginning to appear not only inside WordPress sites but also in the process of building and distributing plugins. The WordPress.org Plugin Handbook describes an MCP server that can help AI-powered development tools consult plugin guidelines, validate readmes, check submission status and submit plugins.
That server supports a developer workflow; it does not bypass review. The Plugin Handbook states, “This is a tool to assist your workflow, not a replacement for the review process.” It also says the usual plugin guidelines apply and submissions receive the same review as submissions made through the web form. This directory-side MCP server is separate from the site-side MCP Adapter: one helps with WordPress.org plugin workflows, while the other can connect site abilities to external agents.
What might the future of WordPress AI look like?
The direction under exploration is a more interoperable WordPress environment in which plugins expose named abilities and agents can discover or call them through defined interfaces. If that develops, site owners could combine capabilities across tools rather than relying only on isolated AI features. The APIs and experiments described by WordPress make that a plausible direction, not a settled forecast or a guarantee that a particular workflow will become a core feature.
Whether that direction is useful in practice depends on choices that remain unresolved across individual sites and products: how permissions are granted, what data is sent to providers, how dependable outputs are, how provider dependence is managed, and which editors and workflows are supported. The current official plugin’s opt-in controls and manual triggers are concrete design choices; they do not settle those broader questions for the whole ecosystem.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
