About
Use an Agentic Avatar to ask questions through a live spoken conversation while watching a video. The avatar responds in the context of the video you’re viewing, helping you better understand its content.
Agentic Avatar responses use the same underlying knowledge sources as text-based Genie responses.
Single-video Genie is available for videos with captions, including:
- VOD
- Audio
- Playlist videos
- Recordings of live sessions
If Genie doesn’t appear, the video may not have captions, the feature may be turned off for your account, or you may not have the required permissions to access Genie.
Start a conversation
1. Open a video (you can do this by clicking the video's thumbnail).
The video page loads with the Genie panel open.
The panel may appear on the right, left, top, or bottom of the video, depending on how your admin set it up. In the example shown here, it appears on the right.
![]()
If the panel is closed, click the Genie icon (stars) in the top-right corner of the player to open it.

If you don’t see the Agentic Avatar option, you can still use Genie by typing your question in the 'Ask me anything' field (see Ask Genie questions about a single video).
The 'Ask me anything' experience shows the following:
- An Agentic Avatar
- A presentation area for supporting content, such as slides or relevant video clips
- A Start conversation button
- The Ask me anything input field
Start a conversation with the avatar
The Start conversation button appears wherever the Agentic Avatar is available. Click Start conversation to begin a live conversation.
Example of Conversational Agent opened from a channel pageThe first time you start a conversation, you’re prompted to allow microphone access.

After allowing access, you can select your preferred microphone and speaker from the device selector on the microphone control.

The experience switches to a full-screen avatar view.
During the conversation
When the avatar loads, it greets you and introduces itself as an agent. If available, it can use basic profile information (e.g. your name) to personalize the conversation.
Speak naturally to the avatar and ask your question.
You can:
- Interrupt the avatar while it’s speaking
- End the conversation at any time
Microphone indicators
When your microphone is active and you’re speaking, a glow appears on the microphone button.

When your microphone is muted, a red indicator appears with a message prompting you to enable your mic.
When interruptions are enabled, a Talk to interrupt message appears while the avatar is speaking.

Use the conversation panel
Click the chat icon in the lower right corner to open the conversation panel.
The conversation panel displays a live transcript of the conversation.
The conversation panel includes:
- Messages you typed
- Questions asked using voice
- The avatar's spoken responses
Use the Ask me anything field at the bottom of the panel to send text messages to the avatar.
Voice and text interactions are combined into a single conversation, allowing you to switch between speaking and typing at any time.
Supporting content
The feature is available only if the agent creator enables it for the relevant agent in Agentic Avatar Studio.
During a conversation, the Agentic Avatar can present supporting content alongside its spoken response to help explain the answer.
Supporting content appears beside the avatar in a side-by-side layout and can include:
- Text such as headings, bullet points, numbered steps, or tables to highlight key information.
- Relevant video clips that automatically play while the avatar explains the answer.
As the conversation continues, the supporting content updates to match the avatar's current response.
Example of supported content in the form of a tableUse page context
If the page context feature is active, the title of the page you're viewing appears above the 'Ask me anything' field in the chat panel.

After you start a conversation, it moves to the lower-left corner of the conversation view.

While page context is active, Conversational Agents answer questions using only the content of the page your viewing.
If you want to ask a question about content beyond the current page, click the X next to the page title, then ask your question.

Conversational Agents will then answer using all your organization's content.
To use the current page context again:
- In the conversation window, click Add page context in the lower left corner.
- If you're in the chat panel, click the + button in the input field and select Add page context.


To learn more about page context, see How page context works in Conversational Agents.
Share your screen
The avatar may suggest sharing your screen when additional visual context would help answer your question or provide guidance.
Screen sharing will only be available if the Share Screen capability is enabled in Agent Studio.
To share your screen:
- Click the screen share icon at the bottom of the screen.
- In the browser dialog, select the desired screen, window, or browser tab, and click Share.
- The shared content is displayed alongside the avatar. The avatar can analyze the content on your screen and respond based on what it sees.


To stop sharing, click the screen share icon again. You can also click the red Stop screen share button that appears when you hover over the shared content.
End the conversation
To end a conversation, click End in the lower right corner to return to the main screen.
After the conversation ends, a summary of the conversation appears in the main panel.
Click Expand conversation to view the full conversation transcript, including all interaction details and sources.
Start a new thread
Click the New thread button at the top right to start a new conversation.

A new thread opens and the avatar screen appears, ready for your next question.
To learn how to share or collaborate on answers, see Understand Conversational Agent responses.
Multilingual support
The avatar can respond in the language you use to ask your question. It automatically detects the language and pulls from matching multilingual sources.
If content isn't available in your preferred language, the avatar returns the most accurate results it can find, even if they're in another language.