In 2026, the AHA Digital History Working Group introduced emerging possibilities for using AI to transform archival collections into structured, analyzable datasets. The On the Books: AI-Assisted Collections project has brought together scholars, librarians and technologists to investigate how Generative AI can support the creation and use of datasets based on archival materials. The 2026 session concluded with a commitment to develop accessible workflows that would enable scholars to apply AI methods to their own research materials and questions.

This roundtable will report on the progress of the project one year later. Scholars will present their contributions to the creation of datasets, how they have used or intend to use the data, and will reflect on the research questions these methods have enabled them to pursue. So far, the project has supported scholarship on the Jim Crow era, African American history, and Chicano/Latino studies. Members from the technical team will present alongside scholars to discuss workflows, infrastructure decisions, and lessons learned from building workflows designed for humanistic inquiry rather than technical expertise.

Rather than focusing solely on technical outputs, the session will foreground the collaborative and iterative process of making AI and other computational tools work for historical scholarship. Discussion will emphasize the role of domain expertise, approaches to working with messy, uncertain, or incomplete archival evidence, and the challenges of aligning computational methods with historians’ research practices. Attendees will leave with concrete examples of adaptable, low-barrier workflows and a clearer sense of where AI can support archival research, along with the firm appreciation that human judgement is always essential.