Skip to main content

Data Source — Files

A Files data source groups documents that your digital employee can use in its responses through semantic search. Supported formats: PDF, Word, Excel, PowerPoint, and text files. Images are not supported.

tip

Group files covering the same topic within a single source. This makes management easier and allows tag-based filtering if you have multiple content scopes.

Managing files​

The configuration page lists all imported files with their title, file name, size, and upload date. Each file can be edited (pencil icon) or deleted (trash icon).

Click "Add a file" to import a new document from your computer.

User access​

The toggle "I want users to be able to view or download information from this data source" controls whether files selected by semantic search are displayed to the user as media cards.

Disable this toggle if the files are only intended to enrich AI responses, without being offered for download to the user.

Media cards​

When a file is selected by semantic search and user access is enabled, it is displayed as a media card containing:

  • A title — the name defined at import, or the file name by default.
  • An optional image accessible from a URL.
  • An optional description. If none is provided, the beginning of the matched passage is shown.
  • A Download button — opens the document. For PDFs, the document opens at the relevant page.
tip

Always provide a clear title and a short description when importing: this improves both the user experience and the quality of semantic search.

tip

For large PDF files, semantic search automatically opens the document at the relevant page, provided the PDF is associated with a reader application on the user's device. This page-level positioning is not available for other formats (Word, Excel, etc.).

Filter responses by tag​

To restrict this source to a specific context (entity, site, department), fill in the "Filter data sources" section at the bottom of the configuration page. See the Overview for details.