The challenge
Retail customers with long or recurring shopping lists often face repetitive catalogue navigation. When they already know the products, quantities, and brands they want, item-by-item search can feel slower than simply stating the list. Arabic- and English-speaking shoppers may also need to express preferences and variants in natural language rather than through filters. Retailers need a more accessible way to capture multi-item intent and turn it into a structured cart without losing customer control over the final order.
Core capabilities
- End-to-end spoken shopping-list capture in Arabic and English. The shopper can speak a complete order rather than search item by item.
- Multi-item quantity, product variant, and preference extraction.The application extracts quantities, sizes, brands, variants and stated preferences from natural-language requests.
- Catalogue search, fuzzy matching, and alternative suggestions. Products and constraints are matched against the retailer’s catalogue, with alternatives proposed when exact matches are unclear or unavailable.
- Automatic cart construction from a single customer request. The interpreted list is assembled into a cart draft for review.
- Clarification prompts for ambiguous, unavailable, or unmatched items. The assistant is designed to ask for clarification instead of silently guessing.
- Reviewable order summary and handoff to the store ordering process. The customer reviews the cart before it moves into the retailer’s ordering flow.
How it works
- 1. Customer speaks the list. A shopper states a long order in Arabic or English, including quantities, sizes, brands, variants, and preferences.
- 2. The application transcribes the request. Speech is converted into text for interpretation.
- 3. Products and constraints are extracted. The system identifies individual items and the conditions attached to them, such as quantity, size, brand, or variant.
- 4. Catalogue matching begins. Each item is matched against the retailer’s catalogue using search and fuzzy matching.
- 5. Ambiguity is handled. Where an item is ambiguous, unavailable, or unmatched, the assistant is designed to prompt for clarification and suggest alternatives.
- 6. A cart is constructed. The application prepares a cart draft from the single customer request.
- 7. The customer reviews the order. A reviewable order summary gives the shopper control over the final basket.
- 8. Handoff to ordering. The reviewed cart is passed to the store ordering process.
Business value
Shoppers can express an entire multi-item order naturally, while the application resolves product details and assembles a cart for confirmation, reducing repetitive item-by-item catalogue navigation.