Libraries, resources and assets

A library groups files for a topic or project. Each uploaded file becomes a resource, and TKP extracts assets from it so that you can inspect and search individual pieces of content.

For example, uploading a PDF report creates a resource for the original file. Text and images extracted from the report become assets linked to that resource. A search can then return a relevant passage instead of only the file's name.

Create and share a library

Open Sources → Libraries, create a library and give it a name. Creating it makes you its Admin. If other people need access, open the library's Members tab and add them with the permissions they need. They must already belong to the workspace; invite them there first if needed. Workspace membership alone does not grant library access; see permissions.

The library contains a folder tree. Its .inbox folder is the initial location for files uploaded without another destination. TKP's AI Librarian can organize processed files into that tree. This organization is separate from extracting content and preparing the file for search.

Upload and wait for processing

Open the library's Resources tab and upload a file. TKP starts ingestion, its background processing of the file. Depending on its format, this can involve extracting text, recognizing text in scanned pages, describing images or transcribing speech. It also prepares the content for search.

The Resources list shows processing status. Wait for Completed or Completed with warnings before expecting the new resource in search results. Warnings mean that processing finished with some limitations. If processing fails or remains incomplete, ask the operator to inspect the resource before uploading it again.

Open the resource to inspect its content and any available pages. The original file remains the source against which you can check extracted text and generated descriptions.

Inspect extracted content

The Assets list shows individual pieces of extracted content, such as text, images and transcript segments. Their source references can identify a page, a time range or a region within an image. This helps you check whether a search result accurately represents the original material.

Pages shows document-page representations where available. Chunks shows the smaller portions of content used for search. You do not need to manage these separately to upload a file; they are useful when inspecting processing results.