Flink Jobs as Agents – Stream Processing for Agentic AI
summary
Claims that Flink jobs become autonomous agents by reading Kafka topics, invoking foundation models, and writing actions to output topics are demonstrated with concrete tooling: Flink SQL gateway (Flip 437 / upcoming Flink 2.1), Flink CEP, Kafka, Iceberg and a conversational platform. Synthetic shadow traffic validates pattern-based fraud detectors. Out‑of‑the‑box AI generation yields ~30–40% successful SQLs, adding schema intelligence raises that to 70–80%, and curated blueprints push to 90–95%. Examples include fraud patterns, supply‑chain disruption detection and predictive maintenance; deployments produce observable output topics and actionable alerts.
AI-written summary · about
chapters (11)
- 0:00Introduction to streaming AI
- 1:04Defining AI agents
- 4:30Flink and stream processing
- 8:07Deploying agents with Flink
- 11:49Complex event processing
- 16:27Fraud detection use cases
- 21:56Schema intelligence architecture
- 27:36AI-assisted SQL generation
- 34:00Agent building tools
- 36:25Advanced agent orchestrations
- 39:50Q&A and closing
related talks
43mEvent-driven Agents with Complex Event Processing in Flink
42mAgentic AI with AWS Streaming
43mArchitecting the Future: Modern Solution Design with Agentic AI
33mAgentic AI – What's Next?
49mAgentic AI Patterns
01h 01mKeynote Panel: Future of AI agents
17mAI TechOps: Data and Infra for AI Foundation to Scale and Manage 1000s of Cohesive Agents
25mFrontiers of Agentic AI