New Plymouth District Council's library network generates millions of circulation transactions each year — checkouts, returns, renewals, holds, and transits — across seven branches and over 1,600 collection categories. Historically, turning this data into timely insight required specialist SQL skills and significant staff time.
In 2025–26, we deployed a natural-language AI analyst built on Snowflake Cortex, enabling library staff to ask plain-English questions of 14.7 million transactions spanning 2018 to today. The system uses a semantic model that maps library business concepts (collections, branches, transaction types) onto a cloud data warehouse, allowing questions like "Which collections are growing in popularity year-on-year?" or "Which branches borrow the most from other branches?" to return accurate, contextualised answers in seconds.
This presentation shares what worked, what surprised us, and what we learned — including the critical importance of semantic modelling, the unexpected queries staff actually asked versus what we anticipated, and how democratising data access shifted conversations from anecdote to evidence. We'll discuss the collaboration between library operations, IT, and Snowflake's AI platform developed by Qrious, and offer practical guidance for other GLAM organisations exploring AI-assisted analytics.
Angela Jowitt has a background as a Digital Library Specialist and now manages the Puke Ariki and Community Libraries. She is leading this project to bring better insights to Libraries through data and digital initiatives. NPDC's library network spans seven branches from Puke Ariki to the New Plymouth Mobile Library. With a focus on making operational data accessible and actionable, she has championed the adoption of AI-powered analytics to support evidence-based collection and service decisions.
Andrew Terwiel is a Data Engineer with New Plymouth District Council's Data & GIS team. Passionate about solving complex data challenges, he helps design and maintain the systems, integrations, and data pipelines that underpin reporting, analytics, and business intelligence across Council. Andrew enjoys turning data into practical outcomes and works collaboratively with colleagues to ensure information is accurate, accessible, and supports better decisions for the community.
Eric Bonhomme is Spark | Qrious’ AI Practice Lead, where he leads the development of enterprise AI solutions that help organisations deploy AI safely, responsibly, and at scale.
He is also completing a Master of Artificial Intelligence by research at the University of Auckland, where his research focuses on Ethical Goal Decomposition, a novel framework for improving the evaluation, explainability, and governance of enterprise AI agents.