MCP App Store
by KeenEthics

Overview

Welcome to the Louis Vuitton App on ChatGPT. Start a conversation to enjoy an inspiring and immersive experience and explore all of the House’s collections. Iconic Bags, Small Leather Goods, Ready-to-Wear, Travel Bags, Gifts, Accessories, Beauty, Perfumes, Shoes, Jewelry, Watches… Dive into the world of Louis Vuitton and ask ChatGPT to discover your favorite Louis Vuitton's creations, from the latest releases to the most timeless pieces.

Tools

search-products

ChatGPT
Explore the Louis Vuitton collection. Handles any query: specific pieces, broad categories, gifts, occasions, or open discovery. The collection covers every Louis Vuitton category — including bags, leather goods, ready-to-wear, shoes, accessories, jewelry, watches, fragrances and perfumes, gifts, and home. A category is best confirmed by searching rather than assumed unavailable. The response is data only: a list of matching pieces (name, SKU, color, material, price, description) plus pagination and the available filters. These raw results are background for curation and aren't shown to the client directly. After each search, either continue researching or prepare the curated selection. Once the research is complete, ALWAYS send one brief, client-facing transition in the client's language before calling render-carousel — for example, "J'ai terminé ma recherche, je vous affiche ma sélection." This transition is a required separate assistant message between the final search-products result and render-carousel. Working with the response: - FIRST SEARCH: use keyword only (plus budget/sort if needed). The available filters in the response are the valid filter options for follow-ups. - REFINEMENT: if the user's intent matches an available filter, apply it via its urlCode. If it doesn't match any filter, a new keyword search without urlCode works better. - ITERATE: several searches (2-3+) tend to give better results than a single one — varying the keyword, applying a urlCode filter, or paging deeper, then comparing results across them. Multi-call exploration surfaces richer options than one call followed by render. - CURATION: pick the best matches for the client's intent, grounded in each piece's description. With zero results, broadening the keyword or relaxing filters and searching again helps. With fewer than 3 results and no specific piece requested, broader terms surface more. render-carousel fits once a good set has emerged across searches. - BUDGET: if the response notes that no pieces fall within the budget, a different category or the most accessible pieces are good options; cheaper pieces are never fabricated. - PRESENT: with 3-5 distinct, relevant pieces, first send the brief transition message, then call render-carousel with their SKU IDs. The transition must be a separate assistant message between the final search-products result and render-carousel. Detailed recommendations follow after the carousel displays, in carousel order. - RE-RENDER: the carousel on screen never updates by itself. Whenever the recommended set changes (narrowed, swapped, or re-curated from earlier results without a new search) render-carousel must be called again with the new SKU set. Recommending in text alone leaves the client looking at the previous, stale carousel. The sweet spot is 3-5 pieces.

Capabilities

Interactive

Example Prompts

Click any prompt to copy it.

Similar Apps