Agora showcased new conversational AI features at the World Artificial Intelligence Conference. These features include voice locking, vision-based intelligence, and interactive avatars.
Agora’s conference showcase gives streaming and video teams a clearer view of the AI features the company is bringing to interactive experiences right now. The mix of voice locking, vision-based intelligence, and interactive avatars points to a product stack aimed at conversational video use cases, not just basic audio. The World Artificial Intelligence Conference provides the timing and venue for this positioning. Watch for whether Agora adds more detail on how voice locking, vision-based intelligence, or interactive avatars will be packaged for video application developers.
Agora has been extending its Conversational AI platform across multiple developer-facing use cases beyond the features shown at the World Artificial Intelligence Conference. The company published a guide for building live voice shopping assistants using its ConvoAI API, which handles real-time audio processing, speech recognition, and LLM orchestration through a single agent-spawning endpoint. The same infrastructure underpins the voice locking and interactive avatar features Agora demonstrated at WAIC, suggesting these capabilities share a common ConvoAI backend rather than operating as isolated products.
Agora's conversational AI engine has also been integrated into game development workflows. The company released a tutorial for building intelligent NPCs in Unity using its ConvoAI Engine, which delivers sub-300ms voice response times even in multiplayer scenarios by running on the same WebRTC infrastructure that powers Agora's live video calls at scale. This positions Agora's voice and avatar stack as applicable to interactive entertainment and gaming, not just customer-facing video applications.
On the model integration side, Agora published a guide for pairing its real-time network with Google's Gemini 3.1 Flash Live Preview to build low-latency multilingual voice agents, noting that its global network powers over 80 billion minutes of audio and video monthly. The guide demonstrates how developers can wire up a production voice agent using Agora's agent quickstart repository, with the same architecture scaling to robotics and conversational commerce use cases. This breadth of model partnerships and deployment targets gives context for how Agora's WAIC announcements around voice locking and vision-based intelligence fit into a broader platform strategy aimed at real-time AI interactions across video, voice, and multimodal surfaces.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source