INFANCIE

Discuss what the customer is looking for to understand their shopping needs. Suited to every part of a shopping conversation: - The user describing what they're looking for - Follow-up questions and back-and-forth dialogue - Feedback on previously shown products (e.g., "too expensive", "too big") - Questions about products, categories, or features The shopping advisor asks thoughtful questions, searches the product catalog (using AI-powered search and matching), and builds an understanding of the customer's needs. Its reply is text only. When it has picked products, the response has has_products: true and a `display` field naming view_product_suggestions and its session_id: that tool shows the picked products to the user as a carousel.

discuss_shopping_needs

Look up a product in this store's catalog by name — the authoritative answer for what it costs, whether it is available, and where to buy it. When the exact product isn't carried, the closest matches come back instead, so the shopper can be told honestly what the store has. Suited to a shopper naming a product or asking whether the store carries something: "do you have the EX40B", "how much is the 15-bar espresso machine", "is the UR25W in stock". Price and availability limits go in the min_price / max_price / in_stock parameters — the query text is matched by meaning, not parsed for numbers. Comparisons, recommendations, and advice on what fits a need belong to discuss_shopping_needs; given this call's session_id, it continues the conversation in the same session.

lookup_product

Record an interaction with the product display shown inside ChatGPT. The display calls this itself — it is not a user- or model-initiated action. It reports one of three interactions: a shopper opening a product's details, expanding the display to fullscreen, or following the link back to the store. No shopper identity is recorded — an event carries the shop, which interaction it was, when it happened, and, where they apply, the product and the version of the display.

record_widget_event

Start a new shopping conversation by clearing the current session. Use this when the customer wants to start over with a completely different shopping need, or when switching to a new customer.

start_new_conversation

View detailed information about ONE specific product: full specifications, dimensions, materials, expert reviews, ratings, pricing across retailers, and personalized fit analysis tailored to the customer's stated needs. Answers when the customer asks about a specific product they have already been shown — for "tell me more about that one", "what are the dimensions / materials / specs / reviews", "is it available in X", or buying intent ("I like that one", "I want to buy it"). It answers for products in the most recent view_product_suggestions carousel. When a session_id is provided, the detail view also generates personalized insights explaining why this product fits the customer's needs.

view_product_information

View personalized product suggestions in an interactive display: the products discuss_shopping_needs picked once its response has has_products: true. With the conversation's session_id, the display shows exactly what the advisor recommended based on the customer's needs. The query parameter is for direct product searches without prior conversation (e.g., user says "show me all sofas").

view_product_suggestions