Nothing new under the sun: Agentic AI and the automated, connected catalog

Abstract

ABSTRACT The emergence of agentic AI is an important technology trend, and its implications for technical services librarianship have the potential to be profound. Agentic AI is introduced and distinguished from generative AI through its capacity to pursue goals, coordinate specialized agents, and carry out multi-step actions with limited human intervention. Current and prospective applications within technical services are examined, including in library service platforms (LSPs) as research assistants, and through the automation of metadata and the potential to enhance retrieval in digital environments. Concerns with regular large language models (LLMs) regarding transparency, bias, hallucinations, governance, accountability, data privacy, and the preservation of professional expertise are compounded in agentic AI environments where agents rely on LLMs. All told, agentic AI represents a continuation of historical patterns of technological change in libraries, and its effective adoption will depend on robust governance practices, ethical oversight, and sustained human expertise.