Most intelligent systems don’t fail because of bad models. They fail because the data arrives late, lacks context, or can’t be trusted enough. Long before AI system response shows up in an application, its limits are already set by the data platform underneath.

This talk looks at intelligent systems from that starting point. Rather than putting all the attention on models, this talk looks at how real-time, event-driven platforms ultimately shape what AI can and cannot do. Talk discusses why batch pipelines fall short for modern use cases and how streaming, stateful architectures allow systems to react as events happen, not hours later. When time and state are treated as first-class concerns, real-time features and low-latency decisions become possible by design.

The talk also highlights the role of “Context” and how Metadata and knowledge graphs help systems understand where data comes from, how it changes, and how pieces fit together. That context makes systems not just faster, but easier to trust and explain.

If you’re building or operating intelligent systems, this session offers a practical view of what really matters before AI ever enters the picture.

https://aiccons.com/keynote_speakers/mr-gajendra-babu-thokala/