Yes. Map a stock_count or WooCommerce stock field into the prompt and instruct the bot to only suggest bottles where the value is above zero. The system prompt can also rank by stock level, recommending the bottle with seven in the cellar over the one with one left. Stock updates flow through on the next request, no re-indexing step.

Yes. SleekAI reads WooCommerce products natively, including price, stock, attributes like grape and region, and the long description. You map fields once in the chatbot config, then the bot can answer 'show me three Burgundies under $80 with at least four bottles in stock' against live data without any extra integration.

Vintage lives on the bottle record as a custom field or product attribute. Pull it into the prompt and the bot can recommend the 2019 over the 2020 of the same wine, or warn when a vintage you list is no longer in stock. For very old bottles, the system prompt can flag rarity and suggest the sommelier confirm condition before opening.

Yes. Display conditions plus role-based prompts mean a logged-in sommelier sees a bot that includes the off-list reserve cellar, while a guest sees only the public list. Multibot also lets you run a 'sommelier mode' bot on a private staff page with different pricing thresholds and bottle pools entirely.

Include your alcohol-free wines, ciders, or pairing teas in the same post type with a flag like category: non_alcoholic. The bot reads the flag and can switch into non-alcoholic recommendation mode when a guest asks. Pregnant guests, sober diners, and designated drivers get a serious pairing rather than a default 'we have sparkling water'.

SleekAI is a one-time WordPress plugin license. You bring your own OpenAI, Anthropic, Google, or OpenRouter API key and pay the model provider directly at standard rates. A typical pairing conversation runs a couple of cents in API costs, far below the per-resolved-conversation fees charged by white-label restaurant chatbot SaaS.

If the guest is logged in, you can persist preferences in user meta (favourite grapes, budget range, dietary restrictions) and inject them into the system prompt on each session. Anonymous guests get a fresh context each visit, which is the right default for a public list. Logged-in members get continuity that feels personal.

Yes. Tasting notes are read as written, so a French-language cellar list answers French queries fluently. For multilingual venues, the system prompt can instruct the model to detect the guest's language and respond in kind, while keeping bottle names and vintages canonical. WPML and Polylang translations are read as separate post objects.