Enterprise Knowledge Search
As the National Archives of Singapore prepared to release newly declassified records from The Albatross File, the National Library Board needed to broaden public access to historically important archives without compromising the accuracy, context, and authority required for those records.
Reported impact
NLB turned a static archival display into an interactive, evidence-grounded public experience that let visitors explore Singapore's independence records through their own questions while preserving trust and archival integrity. The exhibit drew strong public engagement and established a governed model NLB can reuse across future collections.
National Library Board, through the National Archives of Singapore, worked with Accenture to build ChatBook @ The Albatross File on AWS as a governed generative AI archive experience. The implementation began by taking source materials from The Albatross File together with related newspaper articles, oral histories, and video, then standardizing those disparate formats into a structured, searchable knowledge base. That knowledge base was connected to Amazon Bedrock so the application could retrieve only relevant archival evidence when answering visitor questions. Within that controlled workflow, Anthropic foundation models generated natural-language responses bounded by the selected records. Accenture also built the user interface, question interpretation, and evidence-selection logic, plus rules that suppress answers when the archive does not support a clear response. The finished system serves answers instantly on exhibit screens, links each answer back to underlying sources, and provides a reusable model NLB can extend to future archive collections.
Materials from The Albatross File, plus related newspaper articles, oral histories, and video, are standardized into a structured knowledge base on AWS. Visitor questions from ChatBook are processed through Amazon Bedrock, which selects relevant archival evidence and invokes Anthropic foundation models within those boundaries. The system returns answers to the exhibit interface with links to the underlying sources and explanations of their relevance.
"100K visitors generating 90,000+ questions through the ChatBook in 5 months"
"4K bookings before opening"
"Every answer is grounded in the collection, with Anthropic foundation models operating within those boundaries."
"It answers only when a question can be resolved through the archival record."
"NLB faced a defining choice: how to broaden access without compromising the accuracy and authority those archives must uphold."
"The knowledge base was then connected to Amazon Bedrock, enabling ChatBook to draw only from relevant archival records when addressing visitor questions."
"The team transformed materials from The Albatross File, as well as related newspaper articles, oral histories, and video, into a structured, searchable knowledge base"
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
AWS Sample Dec 17, 2024 Read AI Read AI has been rapidly delivering new features that customers want—for example, custom integrations with business applications including Google Meet, Zoom, Slack, Microsoft Teams, Salesforce, Hubspot, Gmail, Outlook,… Around 30 percent of Read AI’s customers are outside the United States, particularly in Africa and South America. AWS Graviton · Amazon EC2 AWS Sample Apr 15, 2024 InfoCert Training and deploying these models was a lengthy, costly process, which limited access to this set of capabilities to larger companies. While InfoCert was searching for a way to streamline such ML integrations, AWS… Additionally, InfoCert expects to reduce financial overhead by 80 percent for startup, development, and… Amazon Bedrock 
Google Cloud Sample Archived · Oct 9, 2026 The State University of New York (SUNY) Now, researchers can skip the technical complexity, giving them more time to focus on what they do best—finding solutions to the world’s biggest problems. 200+ researchers launching more than 80 projects in the first year AI Hypercomputer · Compute Engine · Gemini platform 