About
Conversational Agents build answers using your organization's verified videos and documents.
Sources help you understand where information came from and provide direct access to the content used to generate a response.
Where to find sources
Sources are available for:
- Flashcard responses
- Text responses
- Agentic Avatar conversations
The Sources section includes:
- Thumbnails for each video or document used
- The entry title, and if available, additional info (like season/episode or document type)
- A plus (+) icon if more sources were used than fit in the visible list
Sources are shown per response. In threaded conversations, each response includes its own Sources section.
Sources may not appear for every response, depending on the content and configuration.
For Agentic Avatar conversations, click Expand conversation under the conversation summary to view the conversation transcript and sources.
Supported source types
The Conversational Agent supports the following source types:
- Video entries - Used with spoken dialogue or OCR (on-screen text)
- Document entries - Including PDFs, PowerPoint files (PPTX), and Word documents (DOCX)
Open a source
Click any source tile to open the original entry in a new tab. When available:
- Videos open at the relevant timestamp
- Documents open at the relevant slide or page
- OCR images shown in flashcards represent captured on-screen text
- The Conversational Agent follows your organization's permissions, and users only see and can access content they’re allowed to view.
- To learn how sources appear within responses, see Understand Conversational Agent responses.
