The One About AI Agents (Round II)
From enhancing RAG systems with intelligent filtering and real-time data extraction, to revolutionising call centers through conversational AI, discover how agentic approaches are creating more adaptive and effective AI solutions that respond intelligently to changing contexts and user needs.
More About the Sharings
Traditional RAG systems often struggle to surface the most relevant information for analysts. Qing Ru (Senior Software Engineer, CSIT) will explore how Agentic AI, through a Relevancy Chunk Agent, enriches RAG outputs and dynamically filters information to help analysts focus on critical insights with greater confidence. She will also share how the team experimented with a code-generating agent to extract real-time web data, expanding discovery beyond static datasets. Through these examples, learn how Agentic AI can complement RAG through adaptive and context-aware orchestration, enhancing information discovery even in constrained operational environments. (Technical Level: 200)
Beng Hui Ong (Senior Solutions Architect, Agora) will share more about "Convo AI Empowering Traditional Call Centers – A Digital Revolution in Progress". Hear more about the general architecture and implementation for AI enabled Call Centers to serve multiple functions and purpose for massive scale. The advantage and value for such implementation, and a quick demo of such capability of an AI enabled call center. (Technical Level: 100-200)
Achieving natural, low-latency multimodal synchronization at scale remains a key challenge in real-time video interaction. Mu Shanwei (Head of Algorithms, Akool) will present an end-to-end architecture that integrates sub-frame facial motion capture, audio-driven animation, and adaptive bitrate encoding to handle dynamic network conditions. The session will outline how lightweight neural priors and edge-optimized inference pipelines enable sub-100 ms end-to-end latency with consistent expressive fidelity. It will also include benchmark results, failure-mode analysis, and deployment patterns across video calling, AR/VR, and interactive broadcast scenarios. (Technical Level: 200)
More About the Speakers
Qing Ru is a Senior Software Engineer on the AI Products team at CSIT, leveraging generative AI to tackle critical national security challenges. She builds and integrates systems that sift through large datasets to surface pertinent information, helping analysts work more efficiently and effectively.
Beng Hui is an APAC Senior Solutions Architect at Agora. With over 20 years of experience in streaming and live broadcast solutions, Beng Hui has played a key role in designing and implementing major projects, bringing deep insights into B2B business practices.
Shanwei is Head of Algorithms at Akool. With expertise spanning model training, architecture optimization, and edge inference, his work explores the broader implications of low-latency generative pipelines in reshaping communication, education, and human-AI collaboration at global scale.
Psst....Interested in becoming a speaker for our sessions? Sign up here!