Orderable sits on top of WooCommerce, so orders live in wp_posts with post_type=shop_order and carry Orderable-specific meta like _orderable_order_type (delivery or pickup), _orderable_timeslot, and _orderable_location_id. Menu products are standard WooCommerce products with extra _orderable_* meta for prep time and availability. Locations live as the orderable_locations CPT, with opening hours and delivery zones in postmeta.

That gives owners a real conversational layer over the Orderable dashboard. A restaurant manager asks how many delivery orders came in for tonight's dinner service, the bot reads shop_order rows filtered by _orderable_order_type and the timeslot meta. An owner asks which menu item sold most last week, the bot pulls WooCommerce order_itemmeta and groups by product. A kitchen lead asks which orders are due in the next 15 minutes, the bot returns a sorted list with table number and items.

Multibot lets a separate public menu bot run on the storefront with access only to active WooCommerce products and orderable_locations opening hours, no access to shop_order rows. Display conditions lock the kitchen bot to staff roles. Conversation logging records every prompt and answer inside WordPress.

Orderable restaurants live and die by the dinner rush, and during service the kitchen has no time for filtering admin screens to answer a quick question. The Orderable dashboard is fine for the visual order queue, but every analytical question becomes a date filter, a status filter, and a click trail through WooCommerce orders. How many delivery orders for tonight, what is the average ticket, which menu items are slowing down, those exist as data and live nowhere convenient mid-service.

A plain-English layer over shop_order rows collapses that work. A restaurant owner can ask the question on a phone between table visits and get a real number, not a guess, because the model reads the same rows the kitchen view reads. For chefs and line cooks the win is even sharper.

A quick question about what is due in the next 15 minutes returns a sorted list, not a scroll session through the order queue. A weekly menu review pulls real units-sold numbers from the same order_itemmeta the admin uses. The data was always there, the bottleneck was the interface.

SleekAI keeps that interface in WordPress, scoped to the right role, with every conversation logged for ongoing audit.