AITENCY — Custom AI Systems
All Case Studies
E-Commerce / RetailNational Hardware Retailer

Customer support overwhelmed by technical product compatibility questions across 20,000+ SKUs

The Challenge

A hardware retailer operating both physical stores and an online shop with over 20,000 SKUs — power tools, cutting equipment, accessories, consumables, and spare parts — faced a persistent bottleneck in customer support. Customers regularly asked highly specific technical questions: "Will this diamond cutting disc fit my Makita GA5030 angle grinder?" or "What's the compatible battery for this DeWalt impact driver model?" Support agents needed 5-15 minutes per inquiry to research specifications, cross-reference product databases, and verify compatibility — often getting it wrong, leading to returns and customer frustration.

The Solution

AI-powered product compatibility engine with natural language consultation

AITENCY developed an AI product consultant that ingested the entire product catalogue — specifications, compatibility matrices, manufacturer data sheets, and cross-reference tables — into a structured knowledge base. Customers could ask natural language questions through the website chat, WhatsApp, or in-store kiosks, and receive instant, accurate answers about product compatibility, recommended accessories, and suitable alternatives. The system understood technical specifications, model numbers, and brand-specific naming conventions, and could reason about compatibility even when exact matches weren't in the database by analysing dimensional specifications and interface standards.

Implementation

The project ran over 4 weeks: 1 week of catalogue data extraction, normalisation, and knowledge base construction from multiple supplier feeds; 1 week of AI model fine-tuning and compatibility logic development; 1 week of multi-channel integration (website, WhatsApp, in-store); and 1 week of testing with real customer queries pulled from historical support tickets.

ClaudeGPT-4PythonpgvectorPostgreSQLWhatsApp Business APICustom chat widget

Results

< 10 seconds

Average response time for product compatibility queries, down from 5-15 minutes

94% accuracy

Compatibility recommendations verified correct, exceeding human agent accuracy of ~82%

35% fewer returns

Reduction in product returns attributed to incorrect compatibility purchases

Ongoing Value

The knowledge base automatically updates as new products are added to the catalogue. The system has become the primary pre-sales tool, handling over 400 compatibility queries daily and freeing the support team to focus on complex after-sales issues. Customer satisfaction scores for the online store increased measurably within the first quarter.

Ready to See Where AI Fits in Your Business?

Book a free process audit. We'll analyse your operations, identify the highest-value automation opportunities, and give you a clear, actionable roadmap — with no commitment required.