From Tacit Experience to Retrievable Knowledge: Designing an LLM-Based Interview System for Knowledge Retention

Knowledge management systems often retain explicit documentation but struggle to capture experiential knowledge that remains latent in individual practice. This paper presents a design science study of a large language model-based interview system that externalizes such knowledge and prepares it for retrieval-based reuse in higher education. The artifact uses phase-guided conversations to elicit student experience, extracts user-assessable claims from the elicited knowledge and stores experiential knowledge in a semantically structured form for later retrieval. A formative evaluation with 13 interview sessions produced 440 conversational turns across 21 course components and 173 semantic knowledge statements. Results indicate that the system supports reflective elicitation and claim-level knowledge extraction, while source grounding remains the central challenge for retrieval-based reuse. The study contributes design principles for generative AI-enabled knowledge externalization and provides practical guidance for retaining experiential knowledge across cohorts and knowledge-intensive organizations.