Automotive · Digital Transformation
Conversational product discovery for a global automotive catalog.
The situation
A global automotive manufacturer had its digital presence scattered across dozens of market-specific microsites that didn't share infrastructure. Each region had built its own. Product data was duplicated, drifted out of sync between markets, and customers comparing models across pages were often comparing different versions of the same vehicle.
What we found
The inconsistent information made comparison difficult. Customers also needed help relating specifications to their needs, such as winter driving or room for a growing family. The project needed to address the product data and how people used it to choose a vehicle.
What we built
Two layers. The first was the platform itself: a modular content architecture with a single source of truth for the product catalog, region-specific microsites that pulled from it, and a publishing model that let local teams configure without forking the codebase.
The second was an assistant that sat on top. Customers could describe what they actually wanted in natural language, like a daily driver that handles winter roads, or something that fits a growing family. The assistant mapped those inputs against the catalog and surfaced the right product line, then handed off to dealer locators and configurators. This was early NLP work, pre-LLM, built around intent classification and structured mapping. It connected product discovery with the next steps in the buying process.
What changed
Regional teams shipped market-specific content without re-architecting every release. The product catalog became one place, not dozens of drifting copies. The assistant connected recommendations to dealer locators and configurators so customers could continue the buying process.