Broader adoption of chat reference tools has led to increased access to unstructured data, revealing patron needs and question trends. To utilize this information to the fullest requires careful analysis of the transcripts - from navigating privacy concerns to identifying user themes. This workshop will walk through the process of how we developed a custom-built, offline, AI-powered tool to help with transcript analysis. We begin by comparing our solution to a baseline, human analysis of our text followed by a side by side comparison to pre-built LLMs such as Gemini, to determine which method was most effective.
