Custom Post Type UI stores everything in two wp_options rows: cptui_post_types holds the registered post types with their slug, labels, supports array, rewrite rules, REST visibility, and dashicon, while cptui_taxonomies holds the taxonomies with their hierarchical flag, object types, and URL settings. Both arrays are the schema of your site. Editors know the slugs by heart, visitors do not, and a generic chatbot has no idea any of it exists.

SleekAI's Wizard maps both options into named variables that the chatbot's system message reads at request time. When a visitor asks "do you have a directory of staff," the bot checks the registered post types and replies with the actual staff CPT plus its archive URL. When an editor asks which post types support thumbnails or which taxonomies are hierarchical, the bot quotes the live setting. Add display conditions and a different chatbot can run on each CPT archive with content scoped to that type.

For development teams, the bot doubles as a schema reference. Ask which post types are exposed via REST, which use a custom rewrite slug, or which still have show_in_rest=false blocking a headless build. Conversation logs save every schema question, so the lead developer can see what newcomers struggle to find. Most of those answers used to require opening the CPT UI admin and scanning a long form. The chatbot collapses them into a sentence.

Custom post types are the schema of a content site. Staff, speakers, projects, members, properties, events, courses, and recipes all live in CPTs that editors define for a reason. CPT UI makes that definition a few clicks instead of code, which lowers the cost of creating the right structure.

The chatbot side has lagged. Generic chatbots have no visibility into CPT schema, so they either misroute visitors or invent URLs that do not exist. SleekAI fixes this by reading the same two options CPT UI writes.

The bot answers schema questions accurately. It tells visitors which archive holds the directory they want, quotes the labels editors picked, and points at the real archive URL. Developers and editors get a faster way to verify configuration than opening the CPT UI admin.

Ask the bot which post types support thumbnails, which taxonomies are hierarchical, or which entities are REST-exposed, and the answer comes back in seconds. The conversation log surfaces the schema gaps. Repeated misses on questions like "do you have a directory of clients" show that visitors expect a CPT that the site does not yet register.

That is a content design signal, not a chatbot bug.