Customer describes — AI finds the product
Customers describe in their own words what they need — the AI finds the right item in the catalogue.
LLM + own knowledge base (RAG)The problem
A catalogue with tens of thousands of items is unusable for customers who don’t know the exact name or article number. The result: calls, queries, abandoned orders.
The solution
The customer describes the problem in everyday language (“seal for a dripping tap, 1990s build”). The AI understands the intent and suggests matching items including alternatives — based on the real catalogue, not invented.
How it works
- The product catalogue is turned into a searchable knowledge base.
- The query is understood semantically, not just by keyword.
- Results come solely from the real catalogue (no hallucination).
- Alternatives and accessories are offered alongside.
What changes
Fewer phone queues, more completed orders, happier customers — and the field team is freed from standard search requests.
Honest take
The key is not a giant model but the clean link to the real catalogue (RAG). That is exactly what prevents the dreaded “hallucination” — the AI may only recommend what actually exists.