Development of an AI-based, customised company chatbot for capturing and leveraging explicit and tacit expert knowledge in purchasing

Manufacturing SMEs are increasingly losing valuable purchasing knowledge due to demographic change, employee turnover, and rising process complexity. Around 90% of this know-how is tacit, scattered among individuals and difficult to access. At the same time, employees reportedly spend up to eight hours per week searching for information – a major efficiency drain translating into billions in costs.

The "EnKi‑Bot" research project addresses four key deficits: the manual effort involved in externalizing expert knowledge, slow access to information, the siloed nature of conventional enterprise search systems, and low user acceptance of new software.

At its core is the development of a chatbot that combines a large language model (LLM) with retrieval-augmented generation. The system queries a company-wide knowledge base, initiates dialog-based "interviews" with experts when needed, (semi-)automatically externalizes their tacit knowledge, and stores it after employee validation.

The technology is accompanied by an implementation strategy including training sessions, feedback loops, and data protection measures to build trust. Expected outcomes include more efficient purchasing workflows, better knowledge transfer, and cost savings – ultimately strengthening SMEs and helping to counteract the skilled labor shortage.