From Institutional Memory to Anticipatory Intelligence: Agentic AI in University Knowledge Management

Abstract

Universities face increasing pressure not merely to archive past activity but to anticipate what is coming, as rankings, funding landscapes, and research priorities shift in response to forces no single institution can fully predict. Traditional knowledge management (KM) approaches, designed to preserve and retrieve institutional memory, answer yesterday's questions with yesterday's data. Agentic artificial intelligence (AI), distinguished by goal-directed autonomy, continuous environmental monitoring, and proactive recommendation generation, offers a qualitatively different proposition: a knowledge system that does not wait to be queried. This study investigates how agentic AI reshapes KM and research intelligence practices at the European University of Armenia, and whether those changes produce enhanced strategic foresight. Using an exploratory qualitative case study, data were gathered through semi-structured interviews with twelve purposively selected participants and triangulated against institutional documents. Thematic analysis following Braun and Clarke (2006), informed by the data-structure conventions of the Gioia methodology (Gioia, Corley and Hamilton 2013), was applied across both data sources, with findings organised around five research propositions. Agentic AI strengthens research intelligence by integrating siloed data streams, enabling real-time bibliometric monitoring, and generating predictive scenarios that inform leadership decisions. This intelligence does not directly produce strategic foresight; it must be interpreted and acted upon by human decision-makers, making research intelligence the indispensable intermediary. Gains in knowledge sharing were also evident, although participants cautioned that fluent AI summaries may erode the reflective dialogue through which academic communities validate knowledge. Governance quality, ethical frameworks, and the depth of human–AI collaboration emerged as equally critical conditions, operating as preconditions for value rather than safeguards added after deployment. The practical implication is that investment in agentic AI must be matched by investment in human interpretive capacity and governance infrastructure. The study contributes empirically to KM theory by showing how agentic AI repositions universities from passive custodians of recorded knowledge to active, intelligence-driven institutions, offers context-specific evidence from an emerging post-Soviet university, and provides guidance for responsible AI integration.