Yes. Stock is read live from WooCommerce post meta on each request, so the bot reflects whatever inventory shows at that moment. If the last copy of a 1000 piece title sells out mid-conversation, the next message in that conversation will reflect the change. There is no nightly catalog dump and no separate index to keep in sync.

Yes. The bot reads WooCommerce product attributes natively, so a query like 1000 pieces, Ravensburger, easy difficulty, coastal theme is treated as four filters joined with AND. It returns the titles in your store that match all four, with stock and price, rather than guessing or returning a wide list that ignores half the filters.

Yes. Each brand has its own missing piece policy and intake form, and the bot quotes the right one based on the brand of the puzzle in question. Ravensburger replaces direct, Heye routes through the manufacturer with a longer turnaround, and store-branded titles can be replaced from your own spare stock. Customers get the accurate path the first time.

Yes, as long as age range is stored as a product attribute or in a custom field. The bot reads that meta and pairs it with the difficulty rating and the theme to make a sensible suggestion. A grandparent looking for a 7 year old gets a 100 piece title with large pieces; a buyer for a teen collector gets a 1500 piece title with a complex image.

Cutoff times and the gift wrap option are read per SKU from post meta. A shopper asking whether a puzzle will arrive before Saturday gets a real answer based on the cutoff for the carrier in use that day and the country of delivery. There is no hedging when the bot knows the cutoff has not passed yet.

All conversation logs stay inside your WordPress database. The data path is browser to WordPress to your OpenAI or Anthropic key directly, with no Sleek server in between. For EU stores subject to GDPR this matters because customer messages never reach a third-party platform other than the AI provider you pay for, and you can export or delete logs through the standard WordPress admin.

Yes. When a shopper has settled on a title, you can configure the system prompt to surface accessories from the same category: roll up mats, sorting trays, glue. The bot reads category and tag relationships from the catalog so suggestions track your real merchandising. Light suggestions perform better than aggressive cross-sell, so the prompt is tuned for one or two related items, not a list of five.

Most hosted chatbots run on a vendor server and read your help center, not your catalog. They cannot answer piece count or brand questions accurately because they have no link to wp_postmeta. SleekAI installs inside WordPress, reads the same product data WooCommerce uses, and runs on your own API key. The shift from policy parrot to catalog assistant is what changes the conversation rate for puzzle stores.