[ Report ]
BiblioCommons AI Initiative: Proof of Concept Phase 1 Findings (2026)
6 months. 3 use cases. 4 library partners. Here's what we found.
What responsible AI looks like in a public library.
BiblioCommons partnered with four public libraries and AI consultancy Yonder to test whether AI could meaningfully improve the patron experience, without compromising the values that make libraries trusted.
Four public libraries shaped every stage of this work by sharing their expertise, challenging the findings, and making sure the results reflect what library staff and patrons actually need.
Libraries have always connected people to the right information. AI is changing how people expect to find it.
AI is quickly becoming part of how people discover and interact with information. For libraries, the opportunity is not simply to adopt AI, but to ensure it is used in ways that reflect their values of trust, transparency, and equitable access to knowledge.
This initiative wasn't about rushing a product to market. It was about learning alongside libraries. Over six months, BiblioCommons and four library partners explored three AI use cases, identified real-world opportunities and risks, and established a thoughtful foundation for future AI development.
The question driving the work: can AI improve the patron experience without compromising the trust libraries have built over decades?
Who Should Read this Report?
Library Directors
AI is changing how people discover information. This report explores what a library-led approach looks like, what we learned through hands-on testing, and how libraries can move forward without compromising the values that matter most.
Public Service and Collection Staff
See how AI can help patrons discover more of your library's collections and expertise while keeping staff in control. The report also explores why Comment Summaries is one of the strongest opportunities for an early patron-facing experience.
IT and Systems Leaders
Built on AWS Bedrock using Anthropic Claude models and a multi-agent architecture, Phase 1 offers a practical look at the technical decisions, implementation lessons, and production considerations that emerged along the way.
"I’m particularly hopeful that this tool would be clearly distinguishable from general-purpose AI tools such as Copilot and ChatGPT. It should offer a notably different, purpose-built user experience.”
Founding Partner Library Survey Response
What Library Leaders and Staff Are Asking About AI
(Click each question to reveal the answer.)
Can public libraries build AI tools that genuinely respect patron privacy?
Yes. Throughout this initiative, no patron data was used at any point. All content sources were publicly available. Protecting patron privacy was a core principle from the very beginning, not something added later. As we continue this work, that commitment will remain unchanged.
How is the BiblioCommons AI initiative different from ChatGPT or other general-purpose AI tools?
The BiblioCommons AI initiative explores how AI can work with library content rather than the open web. In Phase 1, the prototypes drew exclusively on publicly available library content and the library catalog, so every recommendation was connected to materials patrons could actually borrow. The goal was to understand how AI can support discovery while reflecting the trust, expertise, and values that libraries bring to their communities.
What does AI-assisted readers' advisory actually look like for a patron?
A patron asks for a book similar to one they loved. AI identifies themes, tone, and reading level, then recommends titles from the library's catalog with an explanation of why each one might be a good fit. It can help patrons get started or continue exploring on their own, while librarians remain the best source for personalized readers' advisory.
Is the BiblioCommons AI initiative available now? How can our library get involved?
Phase 1 was a proof of concept designed to test ideas, learn from library partners, and identify what comes next. Comment Summaries showed the clearest path toward a patron-facing experience and is now moving into structured validation as part of Phase 2. If your library is interested in participating or learning more about Phase 2, we'd love to hear from you. Email info@bibliocommons.com to get involved.
Will AI replace librarians or reduce library staffing?
No. Every use case explored in this initiative was designed to support library staff, not replace them. By helping answer routine questions or surface relevant information more quickly, AI can give staff more time for the advisory, relationship-building, and community work that only people can do.
What library values guided this initiative?
Every decision in this initiative was guided by the same principles: building something patrons can trust, being transparent about how AI works and where it falls short, focusing on genuine patron value rather than novelty, prioritizing accuracy in every response, designing for equity and accessibility from the start, and treating patron privacy as a non-negotiable baseline, not an afterthought.
[The AI Initiative] improves patron independence to find easier information, while connecting them to humans for harder questions."
Founding Partner Library Survey Response
The report covers:
-
Research data on patron search behaviour and the content gap that validated the initiative
-
Full use case findings and assessments for all three use cases
-
Technical architecture overview
-
Direct survey quotes from the founding Partner Libraries
-
Phase 2 exploration areas and the next steps