File-based data libraries and Intelligent Context processing
Key Takeaway
A file-based data library uses Intelligent Context by default. You can't turn Intelligent Context on or off after you save the library, so if a saved library needs it, create a new library with Intelligent Context turned on.
An Agentforce Data Library gives an agent your own content to ground its answers. A file-based data library indexes files you upload, such as PDFs, in Data 360. By default, it applies Intelligent Context, an AI-powered processing step that builds a better search index, especially for tables and infographics. Intelligent Context costs extra credits.
What a data library is
An Agentforce Data Library indexes your content in Data 360 so an agent can retrieve it and ground its responses. Data 360 must be set up before you can use data libraries.
Each data library uses one data source: knowledge articles, file uploads, or a custom retriever. You can't change the data source after you select it, so choose the one that matches where the content lives. A library that uses file uploads is a file-based data library.
A bike-rental shop keeps its assembly guides and safety checklists as PDF files. Because the content lives in files, the admin selects file uploads as the data source and builds a file-based library. A year later, the shop moves those guides into knowledge articles. The admin can't switch the existing library's data source, so the admin builds a new library that uses knowledge articles.