The resources industry operates within an increasingly complex data landscape. Decades of exploration activity have produced vast archives of geological reports, surveys and data sets, yet extracting meaningful insights from these collections remains time-intensive and fragmented. While modern data portals have improved access to individual reports, geoscientists still face the challenge of synthesising insights across thousands of documents. The Geological Survey of Queensland (GSQ) initiated the Digital Librarian Proof-of-Concept to explore how Generative Artificial Intelligence (GenAI) could assist exploration professionals in discovering and synthesising information across large collections of geological reports. Using large language models combined with retrieval-augmented generation techniques, the system enables natural language queries across curated report collections and returns synthesised responses derived from the underlying geological literature. The proof-of-concept demonstrated that GenAI can significantly accelerate data discovery by extracting insights from unstructured exploration reports and presenting them in accessible, contextualised responses. This capability has the potential to shift exploration workflows from manual data searching toward higher-value geological interpretation and decision-making. The project also revealed the important role that this kind of proof-of-concept plays in helping broker organisational change by enabling stakeholders to visualise both the significant benefits and potential risks in AI deployments. These insights highlight both the promise of AI-enabled discovery tools and the practical challenges associated with deploying them in operational environments. This presentation shares the design, outcomes and lessons from the Digital Librarian experiment, and explores how AI may transform geological data discovery while supporting faster, more informed exploration decisions.