It works with any reservation plugin that stores bookings as standard WordPress posts or custom posts, which covers the common ones including Amelia, Bookly, and most native WooCommerce booking plugins. SleekAI reads availability data via postmeta or via a custom endpoint you wire in the admin. For cafes using a third-party booking platform outside WordPress, the bot can hand off to the booking URL after confirming the question rather than answering reservation queries directly.

Yes, that is the recommended setup. A game_library CPT with fields for player count, duration, complexity rating, language dependence, expansions present, and teach-time gives the bot rich structured data on every title. SleekAI reads CPT data natively without any conversion. If your library is in a Google Sheet or Notion today, a one-time import to CPT is usually faster than wiring an external lookup.

Yes. A player asking for "a 90 minute strategy game for four" triggers a query against the library filtered on duration and player count, with a complexity filter inferred from "strategy". The bot returns two or three suggestions from titles in stock, with the slot in the library where they live. This is materially better than a static "browse by category" link on a website that already has too many menus.

It handles the common patterns: party size, day, time window, duration, and whether food is involved. For complex multi-table or private-event inquiries the bot collects the structured fields and either confirms the booking through the reservation plugin or forwards a structured summary to the staff inbox. The point is to absorb the easy 80% so the team can focus on the 20% that genuinely needs a human.

If you maintain a "currently in play" status on library entries the bot will respect it. Most cafes do not because the overhead of updating mid-shift is too high, so the bot defaults to library inventory and notes that availability of any specific copy depends on what is on tables at the moment of the visit. For high-demand titles like Wingspan or Cascadia, that single-copy caveat sets the right expectation.

Yes. The underlying model handles all the languages OpenAI or Anthropic supports, so a cafe in Berlin can take questions in German, English, Turkish, and Spanish without separate setup. The library and event data are language-agnostic. You can pin the bot to a single language with a system prompt instruction if you want all replies to land in one tongue regardless of input.

All conversation logs stay in your WordPress database in a SleekAI table. Nothing is sent to Sleek servers because data flows browser to WordPress to OpenAI directly via your API key. This matters for cafes in the EU under GDPR and for owners who want guest interest signals to stay in-house. Standard WordPress database export covers any DSAR request and any audit workflow.

There is no per-conversation markup from SleekAI. You pay your model provider for the underlying tokens, which for a typical cafe interaction lands in the fractions of a cent on a current-generation OpenAI or Anthropic model. The SleekAI plugin is a one-time purchase plus an optional annual support and update plan. For a cafe doing a few hundred chats a month the model bill is single-digit EUR.