Conversational AGI: The Next Chapter for Real-Time Engagement
Tony ZhaoFounder and CEO, Agora
Panel Discussion
Tony WangCo-founder, Agora
Panel Discussion
Moderator: Bo CuiChief Business Officer, Shengwang
Panelist: Wei ChenFounder, Xieyue Intelligence
Panelist: Zhili LiuCEO, Guanke Technology
Panelist: Guoxiang CaiCo-founder and Head of Ecosystem, Rokid
Panelist: Kan WuCEO, Titan Qiming
Agora Spotlight
Curated by Guanghua Yao and Ting Li; Media partner: Everyone Is a Product Manager
Agora Open Mic: What's New This Year
The latest Shengwang product launches and company updates.
What's New in Shengwang AI
Guanghua YaoHead of AI Product Line, Shengwang
RTE Next: Unified Communications as the Foundation for Real-Time Agent Interaction and Product Launch (tentative)
Xuebin YangShengwang
From Conversation to Outcome: A Conversational AI Engine for Enterprise Voice Agents (tentative)
Yujie MaoShengwang
AI + RTE Use Cases from a Global Perspective (tentative)
Ben WeekesAgora
Smarter Conversations, Connected Experiences: IM 5.0 and the Next Era of Real-Time Engagement
Can WangShengwang
Real-Time Speech, Translation Everywhere: Advances in Shengwang's Transcription and Translation Products
Fan YangShengwang
Stop Reinventing the Wheel: Building a Real-Time Collaborative Meeting Engine (tentative)
Xiangzhong JiShengwang
The Evolution of Cloud Recording: From Storage to a Core Cloud Media Component (tentative)
Rongjun MeiShengwang
Content and Experience Together: Licensed Karaoke Catalogs and Beauty Effects for Social Entertainment (tentative)
Shihao RaoShengwang
Open Connections, Strong Boundaries: Security and Compliance for Real-Time Intelligence (tentative)
Yuan XueShengwang
Real-Time Intelligence for Good (tentative)
Lei QianShengwang
Enterprise Customer Service Agents
Curated by Fan Yang and Ying Li; Media partner: Jiazi Guangnian
The Last Mile of Customer Service Agents: From Conversation to Resolution
Enterprise customer service agents are moving rapidly into production. This forum explores the evolution of Voice Agents, end-to-end business workflows, and how to measure value as agents move from conversation to resolution, conversion, and growth.
Shengwang Helps Agents Communicate and Convert, for Just RMB 2,999 per Month
Zhe HuangCall Agent Product Manager, Shengwang
How to Capture the AI Voice Agent Opportunity in the Large-Model Era
Zhili LiuCEO, Guanke Technology
Voice Agent Technology and Practice in Enterprise Scenarios
Jun XuSenior Technical Expert, Meituan
Self-Evolving Customer Service Agents
Xinhua ShiSenior Architect, Alibaba ATH
From Cascaded to End-to-End: Global Voice Agent Trends in 2026 and Opportunities for ASR x TTS
Ruhuan ZhangSenior AI Architect, MiniMax
Global Practices of AI Outbound Calling
Patrick Ferriter Agora Senior VP Product
Panel Discussion:AI Agents at Work — From Talking to Taking Action
Moderator: Sam Yao Agora Senior Director of Global Product
Panelist: Chi ZhangVice President, Public Cloud Business Unit, Alibaba Cloud Intelligence
Panelist: Zhilong Qian Chairman, Anxiaoxin
Conversational AI Technology
Curated by Lei Xie, Professor and PhD Supervisor at Northwestern Polytechnical University, and Bo Wu, Head of Multimodal AI at Shengwang
Interaction Is Intelligence: Conversational AI and Interactive Models
From end-to-end speech models to new multimodal architectures that listen, speak, reason, and act at once, this forum explores the evolution of conversational AI, full-duplex interaction, and the boundaries of next-generation models.
Audio Intelligence in the Foundation Model Era: From Understanding and Generation to Interaction and Agents
Lei XieProfessor and PhD Supervisor, Northwestern Polytechnical University; Director, ASLP@NPU
Multimodal Models and Real-Time Voice Interaction (tentative)
Zeyuan WangAudio Algorithm Expert, Shengwang; focuses on multimodal models and human-computer interaction
Target Speaker Extraction and Multi-Speaker Recognition in Complex Group Conversations
Ming LiProfessor, The Chinese University of Hong Kong, Shenzhen
Loxi-Speech: Speech Synthesis Built for Real-Time Interaction
Xinsheng WangHead of Soul Loxi and Co-founder, Loxinity
Technical Perspectives on the Next Generation of Interactive Models
Xuerui YangHead of Speech and Interactive Models, StepFun
Panel: Bringing Real-Time Interactive Models into Practice
Moderator: Bo WuHead of Multimodal AI, Shengwang
Panelist: Guangjian XuHead of Audio Technology, Liangyuan Xinchuan
Panelist: Zhifei XieNTU PhD; author of VoiceMem, Mega-ASR, and Mini-Omni
Panelist: Bin YangFounder and CEO, BreezeBlue AI
Panelist: Xinsheng WangHead of Soul Loxi and Co-founder, Loxinity
RTE 101
Curated by Huancheng Bai, RTE Developer Community Evangelist and AI Product Manager, and Zhen Chen, Agora Developer Experience Architect
Voice Agents 101 for Developers
What technical hurdles stand between a demo and a production-ready voice agent? This forum covers the RTE fundamentals developers need most, from audio pipelines and full-duplex conversation to interruption handling and latency optimization.
Annual Use Case Showcase
Curated by the Supersonic Program and RTE Developer Community
RTE Demo Day: New Frontiers in Real-Time Engagement
A showcase of the year's most compelling conversational AI hardware and software. Discover possibilities for the next generation of real-time engagement through emerging ideas and unconventional perspectives.
Social and Entertainment
Curated by Mingya Hu and Ying Li; Media partner: Beluga Global
Beyond Boundaries: From Human Connections to Human-Agent Interaction
Global in-app spending on social apps reached USD 15 billion in 2025, up 16% year over year. Generative AI is extending real-time engagement from person-to-person interaction to human-agent experiences. This forum explores new social entertainment formats, real-time content generation, AI companionship, and global growth.
New Possibilities for AI-Powered Social Entertainment and Full-Stack Capabilities
Sifan CaiHead of Social Entertainment Products, Shengwang
Taking Social Apps Global: Breaking Through with Localized Operations (tentative)
Yuzhi ZhangCTO, HOLLA Group
From Finding People to AI Finding People: A New Growth Channel for Global Social Apps
Chao SuFounder, Duoshou Technology
Technology and Challenges in Real-Time Game Screen Sharing
Shipeng LiCo-founder and CTO, KOOK
Exploring Real-Time, Two-Way Digital Humans for AI Companionship
Yin ZhangCEO, Miaokan Technology
Panel Discussion: Reinventing Sports and Entertainment Experiences with Real-Time Generative AI
Moderator: Sam Yao Agora Senior Director of Global Product
Panelist: He BaiPrincipal Architect, Global Media & Entertainment, AWS
Panelist: Yi LvCo-founder, Moment AI
Panelist: Ben WeekesAgora
Smart Hardware and Embodied AI
Curated by the RTE Community, Dongdong Zuo, Zhijie Li, Xiaodong Feng, and Zikun Zhang; Media partner: Zhixingxing
The Real-Time Evolution of IoT: AI Growth and Open-Source Devices
IoT is expanding into homes, education, and companionship. Real-time engagement and large language models are making previously impossible experiences practical. This forum covers multimodal intelligence, models and developers, chips and devices, global growth, open-source innovation, and new AI hardware.
Opening Remarks
Bo CuiChief Business Officer, Shengwang
Differentiation and Added Value in Intelligent Devices
Zhijie LiHead of IoT Products, Shengwang
On-Device AI for Robotics
Lin HuangProduct Manager, Robotics Product Line, Rockchip
Uniview's New Vision for Home Security Worldwide
Chengyuan JiangVice President, Smart Home Division, Uniview
Intelligent Products for Pets
Weixue GuoCEO, PETKIT
From Speech to Observation: How Multimodality Gives AI Genuine Understanding
Xiaodong FengHead of Physical AI Products, Shengwang
Open-Source Chips
Xueli LvGeneral Manager, Ecosystem and Applications, Beken Corporation
From Open-Source Robot Dogs to Robot Ducks: Bringing Desktop Embodied AI to Everyone Through Vibe Coding
Shinian LiCPO, LuWu Intelligence
Audio AI Technology
Audio AI Technology
Voice AI and Acoustics: Real-Time Speech Engineering in the Real World
From speech recognition and acoustic processing in challenging environments to industry adaptation, personalized recognition, and security verification, this forum focuses on the practical challenges of deploying Voice AI.
Next Generation RTE Technology
Curated by Lianfu Hao, Chief Frontend Architect at Shengwang
Decoding the Next Generation of Real-Time Engagement
Explore the evolution of core RTE technology, from high concurrency, global networks, and quality observability to audio and video codecs, WebRTC, and SDKs.
AI Across the Development Lifecycle of Large Distributed Software
Yong ZouTechnical Architect, Easemob
Education
Curated by Mingya Hu and Ying Li
Online Education Reimagined: Innovation in the AI Era
Online education is entering a new phase of sustained growth as learner needs and teaching models evolve. AI is improving content creation, personalized learning, tutoring, and assessment, while real-time engagement enhances the learning experience. This forum explores practical AI, interactive education, and new paths to growth.
A New Chapter for Education: AI and Full-Stack Real-Time Engagement
Rongjun MeiShengwang
Taking Education Global: Low Latency and Full-Stack Interaction
Xiao LiHead of International Business, PalFish
Developer Tools
Curated by Qianze Zhang, Head of Developer Experience and Platform at Agora
Tools as Productivity: AI Coding, Skills, and Agent Development
AI coding is changing how software is built, while Skills make expertise, tools, and workflows easier to capture and reuse. Agent development is moving beyond simple calls toward richer collaboration and workflows. Practitioners share what they are building, the challenges they face, and the changes that matter in developer tooling.
Long-Horizon Tasks, AI Swarms, and Human-AI Collaboration
Feihu TangHead of Developer Relations, Moonshot AI
Beyond Multimodality: Building Omnimodal AGI with Open Weights
Yuan LiHead of Open Source Technology, MiniMax
From Remembering You to Acting for You: Continuous Memory for Multi-Agent Collaboration
ElliotHead of Open Source Ecosystem, Shanda EverMind
Video AI Technology
Video AI Technology
From Real-Time Video Generation to Interactive World Models
AI video generation is moving from asynchronous workflows to real-time interaction and converging with conversational digital humans and world models. This forum explores real-time video, spatial audio, and codecs for the next generation of immersive experiences.
From Real-Time Perception to Long-Term Memory: Continuous Visual Context for Personal AI (tentative)
Junxiao ShenFounder, Memories AI
From AI Development Boards to Phones: Heterogeneous Real-Time Optimization for Neural Image Codecs
Changcai LaiSenior Engineer, Hangzhou Dianzi University
How Real-Time Interactive Models Unlock New Entertainment Experiences
Jiaxin ShiFounder, Xmax AI
From SageAttention Acceleration to Vidu S1/S2: Video Generation in the Real-Time Interactive Era
Jintao ZhangHead of Streaming Video and Model Inference, Shengshu Technology
Real-Time AI Infrastructure
Real-Time AI Infrastructure
Real-Time Inference and the New AI Infrastructure
As large models expand beyond text into speech, vision, and omnimodal inference, low latency, high concurrency, and streaming orchestration become essential infrastructure challenges. Infrastructure experts explore the foundation for the next generation of real-time intelligence.
End-to-End Engineering for Streaming Multi-Speaker Recognition and Real-Time TTS
Yang GaoHead of AI Infrastructure, Mosi Intelligence
On-Device Real-Time Inference and Streaming Full Duplex: Running Omnimodal Models on a Mac
Tianchi CaiHead of Multimodal Infrastructure, ModelBest; OpenBMB Open Source Lead
Using Agents to Accelerate Multimodal Inference
Ruoyi ZhangSGLang-Omni Maintainer
About iRTE
Founded in 2015 by Agora and the RTE Developer Community, the Real-Time Engagement Conference has been held for eleven consecutive years. What began as a technical exchange for developers has grown into one of the most influential gatherings in audio and video technology. Along the way, it has followed the growth of real-time engagement, from RTC technologies to a broad ecosystem of RTE experiences and industries.
This year, RTE evolves into iRTE, the intelligent Real-Time Engagement Conference: the first industry-wide summit dedicated to real-time engagement between human and AI. Explore how RTE technologies can become the nervous system connecting multimodal models, embodied intelligence and AI agents with the physical world, bringing AI from thinking in the cloud to interacting in real time.