Discussions around intelligent systems tend to center on AI models and algorithms, but that’s not always where problems start. Problems often start much earlier, when data arrives late, lacks completeness, can’t be trusted, or is missing critical context. Long before intelligence shows up in an application, it is shaped by the data platform underneath it.

This talk takes a step back and looks at intelligent systems from a practical angle: the data platforms that enable them. Many systems still rely on batch pipelines, which made sense when decisions could wait. That approach no longer works for many modern use cases. Especially in today’s AI era, getting data faster matters more than ever. This talk explores how streaming, event-driven architectures help systems react in real time rather than hours later. It explains how stateful processing and scalable pipelines keep data flowing continuously, instead of relying on snapshots that quickly fall out of date.

6th International Conference On Computational Intelligence (ICCI2025)