Design of an Artificial Intelligence Based Knowledge Sharing System to Support Electronic Procurement Services

Digital transformation of government procurement involves consistent user support for the Electronic Procurement System. However, knowledge about system utilization remains fragmented across multiple platforms, is not reviewed periodically, and is not managed through a structured knowledge sharing mechanism. This study aims to design and evaluate a knowledge sharing system based on artificial intelligence for electronic procurement support services. The study uses an exploratory mixed method design by integrating Soft Systems Methodology into Design Science Research. Data were collected through interviews, document analysis, questionnaires, and prototype testing. The prototype includes a virtual assistant based on retrieval augmented generation, a centralized knowledge base, source attribution, feedback fitur, question logs, a monitoring dashboard, and knowledge governance functions. Evaluation using twenty answerable questions produced faithfulness of 1.000, answer relevancy of 0.819, context precision of 1.000, and context recall of 0.934. Blackbox testing confirmed that the main functional requirements operated as designed. The System Usability Scale score from 43 respondents was 79.24, indicating good usability, while 102 respondents rated all knowledge management process dimensions in the high category. The results show that integrating organizational problem structuring, knowledge governance, and retrieval augmented generation can provide a credible foundation for improving knowledge access and user support in digital public procurement.