Challenge
Many customers configure their vehicle without discovering relevant optional equipment. The group wanted personalised recommendations - but as a service that works for several brands, fits into existing sales applications and does not depend on a single AI vendor.
Solution
- AI-based recommendation system that identifies relevant optional equipment from existing configurations.
- Vector search with similarity logic - recommendations only above a defined threshold, so they actually fit.
- Scalable recommendation service connected to existing sales applications, running in the customer’s cloud environment.
- Proof of concept with the client’s business and IT teams.
- Evaluation of an own language model to secure independence from third-party providers.
Result
- Personalised suggestions as a lever for conversion and customer satisfaction.
- Usable across brands instead of one isolated solution per brand.
- An architecture that allows switching models - sovereignty as a design principle.
* Project: we owned the delivery · ● in operation: we run the solution today
