In the era of rapidly advancing large language models and Agentic AI applications, the integration of big data technologies—particularly real-time processing engines like Apache Flink—is a critical topic for data infrastructure professionals.

Agentic AI applications can be categorized into two types based on their triggering mechanisms: human-request driven agents and event-driven agents. ● Human-request driven agents often require contextual information to fulfill requests, and Flink’s real-time data processing capabilities ensure the timely delivery of updated contextual information. ● In contrast, event-driven agents are triggered by machine-generated events (such as transactions, web clicks, or sensor data), which typically occur at a significantly higher frequency than human requests.

As AI continues to advance and achieve industry-scale deployment, we anticipate that the majority of future Agentic AI workloads will originate from event-driven agents. In response, we are developing a new Event-Driven Agent Framework based on Apache Flink.

This presentation will cover: ● The definition and application scenarios of event-driven agents ● The technical challenges of building event-driven agents ● Why Apache Flink is well-suited for building event-driven agents ● The design principles and current development status of the framework

Speakers:

Xintong is an Apache Flink PMC member. He is also a Staff Software Engineer and leads the Flink Java Engine Team at Alibaba Cloud. Prior to that, he received a Ph.D. degree in computer science from Peking University.