JetReviews stores reviews as WordPress comments with custom commentmeta keys for the overall rating, each criterion score, pros, cons, and a short review summary. Aggregate averages and counts are cached in the related post's postmeta. SleekAI's data sources map all of that into named variables, so a question like "how do people rate the comfort" returns the real average from the criteria meta, not a guess about the product category.

Display conditions scope each bot by post type, taxonomy term, user role, logged-in state, and URL pattern. That means the product-page bot can quote the top three pros and cons from real reviews on that specific item, the category-page bot can compare two items by criteria averages, and an admin bot can answer questions like "which products lost stars in the last month" without anyone digging through the comments table.

Every chat is logged inside WordPress with the model name, token usage, and origin page. Pair that with the JS API or PHP embed and the bot can drop a summary into any Bricks or Elementor section, or surface in a floating widget right next to the review block.

Reviews are the single most read part of any product page, and they exist precisely because copy on the page is not enough. Visitors want to know what real owners think about comfort, durability, fit, and warranty. A generic chatbot has none of that data.

It quotes invented star averages, makes up review counts, and contradicts the rating block right next to it. The visitor notices, trust collapses, and the bot becomes a liability rather than a sales aid. JetReviews already structures this data well.

Overall ratings, per-criterion scores, pros, cons, and the short summary all sit in WordPress comment meta, with aggregate averages cached on the related post. SleekAI maps those fields into the chatbot prompt with point and click data sources, so the bot quotes the same numbers the rating block shows, and can quote real reviewer phrases when context helps. Per-bot display conditions then let you run a product-page bot, a category-page comparison bot, and an admin-only sentiment bot from the same install, each scoped exactly where its data is most useful, and each running on a key the merchant controls.