AI Shopping Assistant for a Luxury Appliance Retailer

Research question

Can conversational discovery improve product visibility and purchase confidence?

The retailer carried a large catalogue of appliances and specialist devices. Customers could not easily discover the full range, and some products remained in stock without sales.

Customers also lacked confidence when checking dimensions and specifications. Resolving these questions required a customer-service call, adding delay and support workload.

Working hypothesis: allowing customers to describe their needs and ask product-specific questions would improve discovery and reduce uncertainty before purchase.

Intervention

ERP-connected conversational shopping assistant

I defined the commerce concept and software architecture with the company's internal AI engineer. The assistant used the retailer's ERP product data to answer natural-language questions and recommend relevant products.

1. 01
   Product discovery

   Interpret customer needs expressed in natural language.

2. 02
   Product information

   Return specifications, dimensions and practical suitability information.

3. 03
   Access points

   Expose the assistant through a prominent floating action button on web and mobile.

4. 04
   Architecture

   Fine-tune a custom assistant to reduce third-party model dependency and control operating costs.

Observed data

Commercial result

The reviewed logs showed that customers encountered the products through assistant recommendations before purchasing them.

Interpretation

Discoverability was a material constraint.

The observed change indicates that the products' previous lack of sales did not necessarily reflect a lack of demand. Customers were more likely to purchase once the catalogue could be explored through questions about need, fit and specifications.

For considered purchases, conversational access to product knowledge can complement filters and category pages while reducing reliance on customer service.