Ask the Library
The Chatalogue
This chat feature is best for questions rather than catalogue search. Ask it something like “what is the state of drinking water on Bowen Island?” and it will read across multiple documents to give you an overall answer with linked references and ideas for followup questions.
The chat considers the content of the conversation as it unfolds. So if you’ve started talking about two specific documents, it may continue to reference them even when you want to shift to a different or wider subject. In that case, it’s a good idea to clear the chat with the eraser button and start fresh.
As with any AI, rather than relying on it entirely for important research it’s best to check the cited sources to ensure accuracy.
Ask the Library is a RAG (retrieval-augmented generative) system with access to all the library docs digitized in a special format called ’embeddings’. Its superpower is that it can search by meaning rather than just text matching, and it sources docs in the collection to compose its commentary. Claude Sonnet is the current chat model at work.
The library catalog, which does not use AI, is the right discovery tool when you want exact, exhaustive search results; browsing by subject category, filtering by document type (reports, bylaws, studies), narrowing by date range, or finding every document associated with a particular author or organization. It gives you a complete, filtered list of matching records with direct links to the source PDFs. Use it when you need to be thorough, or when you already know what you’re seeking. For thorough research, use both tools together.
Ask the Library lets you compare and contrast many documents at once, and is good for broad research. Another thing to try is downloading the text or PDF versions of the source documents and popping them into your own AI. You may find the text versions cheaper and faster to process. It’s possible to install free, open source AI models on your own device, and analyze the documents on your computer rather than a server farm.