German automaker Audi is aiming to boost productivity and decision-making among its employees with the help of an artificial intelligence-powered chatbot developed by tech consultancy Storm Reply.
The chatbot, which is built on Amazon’s SageMaker platform, uses a technique called retrieval-augmented generation (RAG) to quickly retrieve answers to employees’ questions from Audi’s internal documentation and data sources.
If the chatbot cannot find an answer with high confidence, it will communicate this to the employee rather than attempt to speculate an incorrect response.
Storm Reply said the chatbot architecture was developed with security as a priority, residing in a private cloud environment to ensure data privacy.
The system draws information from sources such as Confluence, Notion and PDFs, automatically keeping Audi’s internal knowledge base current.
Michael Pawelke, Audi’s product owner of AWS foundation services, said the firm views the chatbot as accelerating its shift to cloud-based tools. “It lowers the barriers to knowledge discovery for our internal stakeholders and should require little additional effort to extend to other use cases,” he remarked.
Storm Reply CTO Filippo Rizzante said prompting techniques can be used to optimize the chatbot’s outputs without altering the underlying AI model, enabling Audi to refine responses over time.