AI Actions in Realtime

Samuel Su
on February 24, 2025Introducing AI Actions in Realtime - We've expanded our AI actions capability to support realtime voice-to-voice chat interactions. Your AI agent can now not only answer questions based on your knowledge base but also perform customized AI actions and display results directly in the realtime chat interface.
Background
Previously, AI actions were only available in text chat mode. Voice-to-voice conversations couldn't utilize features like lazy live chat, image and document generation, or custom API calls. With this update, we've bridged this gap by bringing full AI actions support to the realtime voice chat.
Key Features
Document and Image Generation
Users can now request document or image generation by talking to the AI agent. The AI agent will process these requests using either the image/document generation API or the AI PDF action, with results displayed directly in the chat interface.

Search and Button Actions
The AI agent can now execute search queries and button actions during voice conversations, with results seamlessly integrated into the chat interface.

Background Actions
Background processes like lazy chat actions and lead collection are also supported in voice conversations. While these actions don't display visible results in the interface, they run silently and influence the AI's responses accordingly.