Toolset Types stores its schema in three wp_options entries: wpcf-custom-types for registered post types, wpcf-custom-taxonomies for taxonomies, and wpcf-fields for the field groups attached to them. The field values themselves land in wp_postmeta with a wpcf- prefix on each meta key. Field groups can attach by post type, by taxonomy term, or by template. The result is a structured content store with custom data on every CPT.

Generic chatbots cannot read any of this. SleekAI's Wizard maps each option and the relevant wpcf- meta keys into named variables, so the bot can answer questions that reference real fields. A property site can return the bedroom count and price for a listing. A directory site can return the staff member's department and start date. A recipe site can return prep time and servings. The bot quotes values from wp_postmeta directly, with no parallel knowledge base.

For sites with deep custom-field libraries, push the field values plus titles plus permalinks into an OpenAI Files vector store so the model retrieves only the matching posts per query. Conversation logs save the question, the matched fields, and the post URL. Editors and developers see which fields visitors actually ask about. That signal is more useful than a heat map. Repeated questions on a field that has no surface UI mean the field deserves a more visible position on the template.

Toolset Types brought structured content to WordPress before that became a standard expectation. Sites built on Toolset usually have rich custom fields per post type, careful taxonomies, and templated front ends powered by Views. The structure is the product.

The chatbot side has lagged behind. Generic bots have no way to read the wpcf- meta or the Toolset options, so they answer in vague terms or invent specs. SleekAI closes the gap.

The bot reads the same wp_postmeta values that Views renders, with the same field-group conditions. Visitors get the actual price, the actual square footage, the actual open-house schedule. Editors get a faster way to verify a field across many posts than scrolling the admin list.

The conversation log surfaces field-level demand. Some fields will already sit in the headline of the Views template because the team understood they matter. Others will sit in a footer or an expandable section that nobody opens.

The log shows which fields visitors ask about anyway. That is the strongest signal for a template revision, stronger than a heat map, because the question is explicit. Compared to building a JSON catalog of the schema, this approach uses the storage Toolset already maintains.

No drift, no rebuilds.