On November 26th, Peking Time, Agora, Inc., the parent company of Agora, Inc. (API.US), released the financial report for the third quarter of 2024. The financial report shows that in the third quarter, Agora, Inc. achieved a total revenue of 31.57 million USD, with Agora focusing its business in the Chinese market generating a Q3 revenue of 0.113 billion RMB. Agora focusing on non-Chinese markets achieved a Q3 revenue of 15.7 million USD.

In terms of cash reserves, the financial report shows that as of September 30, 2024, Agora, Inc. had 0.363 billion USD in cash, cash equivalents, bank deposits, and financial products.

"Recently, Agora launched the Conversational AI SDK and partnered with OpenAI Realtime API to enable developers to integrate voice-driven conversational AI capabilities into any application." Zhao Bin, the founder and CEO of Agora, Inc., commented, "We believe that multimodal AI capable of natural conversations with humans will be widely used in various scenarios such as customer service, education, and health. We aim to become the key infrastructure for conversational AI."

"To support this vision, we have made some organizational adjustments aimed at building a more efficient and flexible organization, allowing us to seize market opportunities more agilely and develop the next generation of real-time interactive technology for generative AI era, consolidating our leadership position in the real-time interactive field." said Zhao Bin.

RTE is a key part of AI Infra, and Agora released the panoramic view of RTE+AI capabilities.

At the 10th Real-time Internet Conference held on October 25, Agora officially released the panoramic view of RTE+AI capabilities, presenting clearly the current technical capabilities, architectural solutions, and application scenarios of combining RTE with AI from five dimensions: real-time AI infrastructure, RTE+AI ecosystem capabilities, Agora AI Agent, real-time multimodal conversational AI solutions, and RTE+AI application scenarios. RTE has become a key part of the AI infrastructure in the generative AI era, and the innovative scenarios brought by the combination of RTE and generative AI will be the theme of the next decade.

Agora was the first in the industry to fully integrate AI algorithms into the RTC technology stack a few years ago, aiming to enhance audio and video experiences as well as QoE. In the audio field, Agora developed the new generation of real-time audio technology with the intelligent engine "Fengming AI Engine", featuring AI noise reduction, AI echo cancellation, spatial audio, and other capabilities to bring a purer and more immersive audio experience to real-time interactive scenarios. In the video field, Agora's "Real-time HD · Super Picture Quality" solution not only significantly improves video quality, but also enhances the experience in areas such as AI beauty, virtual backgrounds, and AI-driven facial expressions and actions.

Generative AI is evolving deeply in two directions: multimodal and multi-step reasoning. Among them, multi-step reasoning is critical for lowering latency and ensuring stable reliability. Agora pioneered the largest real-time audio and video network to date - Software Defined Real-Time Network (SD-RTN). SD-RTN will provide crucial assurance for distributed multi-step reasoning in latency control. Its guarantee of transmission quality and stable reliable support will also become indispensable network infrastructure in the era of generative AI.

Building upon the infrastructure, this quarter Agora has deepened its collaboration with multiple clients in the AIGC scene to explore real-time multimodal conversational AI solutions. This solution, with voice as its core, supports video extensions and enables combined inputs and outputs of text / audio / images / video, creating a real and natural AI voice interaction experience with rich functionality.

The real-time multimodal conversational AI solution offers four major advantages: optimizing interaction latency to less than 500ms; supporting AI noise reduction, background voice filtering for purer voice conversations; supporting interruptibility and hyper-realistic voice synthesis for smarter and more realistic conversations; as well as flexible and scalable AI Agent architecture and other functionalities, enabling developers and businesses to quickly build AI real-time voice dialogue services tailored to their own business scenarios.

Innovative AI+ meeting collaboration, smart hardware scenes, interactive experiences comprehensively upgraded.

As a pioneer in the global real-time interactive cloud services industry, Agora has been exploring the enhancement of experiences and scene innovation brought by the combination of generative AI and RTE across various industries. This quarter, Agora officially launched the aPaaS product "Dynamic Meeting", reshaping the meeting collaboration experience by integrating RTE technology with generative AI capabilities, providing customers with a low-code video conferencing platform. It adopts a multi-cloud, multi-platform design, offers SFU+MCU hybrid architecture deployment capabilities, and supports flexible deployment and extension, fully catering to customer meeting, collaboration, and scheduling needs.

RTE technology ensures low latency, high-quality audio and video transmission, making video conferencing communication more real-time and smooth; while generative AI provides intelligent auxiliary functions through analysis and generation of meeting content, such as real-time subtitles, AI meeting summaries, etc. With the support of these two major technologies, meetings have been upgraded from traditional information transmission tools to intelligent collaboration platforms, effectively improving meeting efficiency and user experience.

In addition, based on the real-time multimodal conversational AI solution, Agora has explored an AI x IoT intelligent hardware solution tailored for the IoT industry. This solution can quickly achieve large-scale model access on low-power, low-compute chips, with characteristics of low-latency real-time interaction and low-cost flexible adaptation, and build a real and natural AI voice interaction experience in intelligent hardware scenarios through rich functionalities.

Currently, multimodal large-scale models are mainly deployed in intelligent hardware scenarios, with wearable devices such as smart glasses, smart watches, and smart earphones as the main devices. Under the trend of generative AI, there have also appeared a series of scenarios including intelligent companion robots, smart plush toys for children, smart rings, bringing AI voice interaction experiences under different hardware terminals.

RTE products are becoming more specialized, launching Status Page and VoIP CallKit this quarter.

This quarter, Agora's RTE products are moving towards a more specialized direction. In his speech at RTE 2024, Zhao Bin mentioned that Agora has launched a real-time interactive Status Page, providing telecom-grade quality assurance and the ability to update quality assurance status at the minute level. Additionally, for experiences such as lagging and delays in real-time interactive dialogues, it provides more detailed and clear indicators to help developers and customers better perceive the status and capabilities of real-time interactive services.

To address the common issue of low user connection rates in the general IPC scenario, Agora recently launched the mini-program VoIP CallKit, based on the Agora RTSA audio and video real-time call service, which supports end-to-end calling functions for WeChat mini programs. It can be applied in various IoT scenarios such as screen-equipped IPCs, smart door locks, companion robots, smart visual desk lamps, smart watches, smart screen speakers, etc. By integrating VoIP CallKit in these scenarios, it efficiently, quickly, and accurately achieves one-to-one real-time audio and video calls between smart hardware and WeChat mini programs.

In August this year, the launch of China's first 3A game 'Black Myth: Wukong' has led to explosive growth in the game streaming business. In order to attract more viewers, the video quality during live streaming has become a major concern for both the live streaming platforms and streamers. Agora's game live streaming solution, through core technology optimizations on the collection and encoding side, ensures real-time video encoding during game screen sharing, with minimal performance consumption and highest encoding efficiency, ultimately achieving a game live streaming interactive experience with no frame drops, no lag, low latency, and ultra-high definition. It truly brings the cinematic 4K 60fps viewing experience to the daily game live streaming of ordinary players.

Published by Agora Research Institute, the new book 'Understanding Real-time Interaction' has been officially released.

Agora celebrates its 10th anniversary this year. Agora has always had a mission of "helping people overcome distances for real-time interaction, as if gathering together", committed to enhancing people's real-time interaction experience through high-quality real-time audio and video technology services, empowering industries such as social, education, finance, medical, etc., and promoting economic and social development. Agora has a responsibility and obligation to promote the popularization of real-time interaction.

In celebration of its 10th anniversary, Agora has released the book "Understanding Real-Time Interaction: In-Depth Analysis of Audio and Video Technology, Scenarios, and Data". Published by Machinery Industry Press, compiled by Agora Research Institute, the book was officially launched on August 27th, making it the first systematic technical popular science book in the industry to introduce real-time interaction.

"Understanding Real-Time Interaction" provides a comprehensive and systematic explanation of the real-time interaction industry from multiple dimensions such as development history, technical principles, application scenarios, and big data observations. Agora hopes that readers can deepen their understanding of real-time interaction, acquire knowledge related to real-time interaction, and encourage more people to join this industry together to drive industry progress.