Product Recommenderv1.0.0

A customer asks for a birthday gift for their mother with a budget of ¥300. The system extracts signals (recipient, occasion, budget) and filters products accordingly. It returns a curated list of 3 suitable items with a gift-specific presentation, including wrapping info if available.

A customer buys a dress and asks 'what goes with this?'. The system identifies the dress as anchor, then filters for complementary items like accessories or shoes tagged with '搭配' or '配套'. It presents a complete set with total price, enabling cross-sell.

After seeing recommendations, a customer says 'a bit expensive, any cheaper?'. The system re-filters with a lower budget (e.g., 30% reduction) and presents alternatives. If none exist, it explains the value of original recommendations without apologizing.

A customer mentions two specific products and asks 'which is better?'. The system fetches both from the knowledge base, builds a comparison table covering price, specs, suitability, and gives a clear recommendation with reasoning. This reduces decision paralysis.

When no products pass filters (e.g., budget too low), the system honestly informs the customer, suggests the closest higher-priced option as an upsell, and offers to notify when a matching product arrives. It only upsells once per conversation.

By recommending complementary products (outfit pairing) or higher-value upgrades (upsell within budget), the system increases average order value. Revenue comes from commission per additional sale.

Offer a premium subscription where customers get unlimited personalized recommendations, gift ideas, and comparison assistance. The skill acts as a 24/7 shopping concierge.

When a customer asks for a product not in inventory, the system can recommend partner brands via affiliate links. Revenue per referral or per sale.