Getting Started with AI-Kit for WordPress

Start with the workflow

Use the simplest execution path that meets the project need.

01 · Setup

Where do you configure and test AI-Kit?

After installation, the main settings area lives under the SmartCloud menu in WordPress admin. Review the defaults first, then use diagnostics to see which on-device capabilities are actually available in the current browser and device.

Availability can differ between environments. Treat diagnostics as a runtime check rather than assuming that every editor or visitor has the same browser support.

• Open AI-Kit Settings to review global behavior and defaults.
• Use Diagnostics before troubleshooting an unavailable feature.
• Connect the site to a WP Suite workspace only when Pro configuration is required.

This keeps the free local workflow independent while making the optional platform and backend paths explicit.

02 · Editor

What can authors do inside Gutenberg?

AI-Kit adds editor tools without replacing the block editor. The sidebar can help generate post metadata or draft text from a topic and instructions, while inline toolbar actions can proofread, rewrite, or translate editable text blocks.

Image-like blocks can expose an SEO metadata panel, so authors can prepare useful alt text, titles, captions, and descriptions while working on the page instead of returning to a separate workflow.

• Generate a draft, review it, then apply it to the post or block.
• Use language and tone controls as editorial constraints, not as substitutes for review.
• Keep accessibility intent in mind when accepting generated alt text.

The value is not automatic publishing. It is reducing repetitive editing work while leaving the author in control.

03 · Media

How does AI-Kit support image metadata at scale?

The same metadata workflow is available in the Media Library, including attachment detail views, edit screens, and bulk-oriented review flows. This is useful when a site has many images with missing or inconsistent metadata.

Generated fields should still be checked against the image and the surrounding page. Alt text in particular should describe the information that matters in context, not merely list visible objects.

• Preview generated metadata before saving it.
• Review bulk items one by one when context differs across pages.
• Do not duplicate nearby captions or headings mechanically.

Used carefully, the tool helps teams improve consistency without turning metadata into an unattended batch process.

04 · Frontend and backend

When do frontend AI features and an AWS backend belong in the design?

Pro includes frontend AI Feature experiences and DocSearch surfaces through Gutenberg, shortcodes, Elementor widgets, and a small JavaScript API. These are appropriate when visitors need controlled writing actions or knowledge-base research directly on the site.

An optional backend becomes useful when on-device support is unavailable, when behavior must be consistent across browsers, or when answers need retrieval from approved WordPress knowledge-base content. The endpoint can run in the customer’s AWS account rather than through a shared AI SaaS runtime.

• Use frontend AI only for a clearly defined visitor task.
• Use DocSearch when answers should be grounded in managed source content.
• Choose authentication, reCAPTCHA, and API exposure according to the route’s risk.

The architecture can stay local-first while adding a governed backend path only where the project justifies it.