Libraries, resources & assets

Libraries

A library is a folder tree that holds documents and media for one purpose or project. Creating a library makes you its Admin (see Permissions & abilities). Add other people to it explicitly from the library's Members tab if they need access.

Every library starts with an .inbox folder: drop files there and TKP's ingestion pipeline picks them up, or an agent can organize them into the rest of the tree automatically.

Resources

A resource is one uploaded file: a PDF, an Office document, an image, an audio or video file. Uploading one starts an ingestion run: the file is parsed, its content (text, tables, images, transcripts) is extracted, described, chunked, and embedded, after which the resource becomes searchable. The Resources list shows each one's ingestion status; opening a resource shows its parsed content, pages, and knowledge graph.

Assets

An asset is one extracted unit of content inside a resource: a paragraph of text, a table, an image, a video shot, a transcript segment. Assets are what actually get chunked and embedded for search; the Assets list is mostly useful for inspecting exactly what TKP extracted from a given resource, and where in the original file it came from (its page, timestamp, or bounding box, the basis for citations).