Mobility
ANWB
Knowledge assistant for member services
Agents type a question and get the right passage from the knowledge base back, with the source attached. Saves a lot of clicking around and second-guessing.
Results achieved
Faster call handling
Consistent answers
Less manual searching
The challenge
Member service agents spent too much time searching the internal knowledge system. They had to guess search terms, read through multiple documents, and often ask colleagues for help to find the right instructions.
Our approach
- 1.Map common questions to existing data sources.
- 2.Build a RAG pipeline on top of the current knowledge base.
- 3.Add source links to every answer for easy verification.
- 4.Refine the system based on feedback from the service team.
The solution
A search assistant that lets staff ask questions in plain text. It retrieves the exact passage from the knowledge base along with the source link.
PythonRAGLLMsVector search