Agora Launches its Conversational AI Engine for Developers

Agora, a platform for real-time engagement APIs, launched its conversational AI engine, to enable developers to create interactive voice experiences with any AI model. It is now available in public beta.

The company claims the conversational AI engine is designed to provide ultra-low latency responses and superior voice processing, resulting in a seamless AI voice experience. 

Agora states that the technology provides developers with the flexibility to use any AI model, whether it is custom-built or from leading LLM providers, and includes support for any text-to-speech (TTS) voice solution for their AI agent. 

Furthermore, it is built to optimise real-time conversation flow, which enables it to give faster, low-latency responses while also pausing for interruptions in real time for a natural dialogue experience.

“Most AI models don’t yet offer the ability to interact via voice, and those that do haven’t optimised the experience,” said Tony Zhao, CEO of Agora. “Agora’s solution will help drive the adoption of conversational AI by enabling seamless, natural voice interactions. With effortless integration and real-time, voice-driven experiences that have near-zero lag, we are removing barriers and unlocking new possibilities for businesses and end users alike.”

The company further explained that the conversational AI engine has built-in background noise suppression, AI-driven acoustic algorithms, and real-time speech-to-text (STT). It is powered by Agorda’s global software-designed real-time network, which helps the engine dynamically handle packet loss and reduce latency.

Agora further states that it is built on the TEN framework, an Agora-supported community project for building conversational AI.

The company is also partnering with Oracle as a key AI infrastructure provider, leveraging the Oracle Cloud Infrastructure (OCI) solution to optimise performance, security, and efficiency.

“Agora’s Conversational AI Engine solves two of the biggest hurdles in human-to-AI voice conversation—latency and wireless last mile challenges,” said Hua Jiang, alliance & channel director, Oracle APAC. 

In the near future, the company plans to extend the capabilities of the conversational AI engine to its app builder product, providing a no-code approach.