AI & knowledge
AI chatbot for sales and product documentation
- Area
- Digital products
- Year
- 2024
A RAG chatbot helps sales staff prepare questionnaire responses and helps customers find clear information in extensive documentation.
Solution goal150,000queries per month
Solution goal< €2,000monthly running cost
Solution goal7%sales staff time saved
Starting point
What needed to change
Preparing the material meant going through the documentation again and again. Besides the quality of the answers, the solution also had to keep the cost of running AI under control.
Our responsibility
How we approached the solution
Elasticsearch first retrieves the relevant context. The language model then summarises it into a uniform format and returns an answer including the sources used.
Part of the delivery
A complete scope
- Context retrieval
- RAG pipeline
- Source citations
- Operating cost control
Impact
What the customer gained
- The salesperson receives material prepared from the relevant documents.
- The end user can trace the sources of an answer.
- The architecture separates retrieving the material from summarising it.
Technologies and approaches
- AI
- RAG
- Elasticsearch
- Cloud