India, April 14 -- At Confluent Data Streaming World Tour 2026, Andrew Sellers made a grounded point about enterprise Artificial Intelligence AI. Many projects do not fail because the model is weak. They fail because the data behind the model is late, fragmented, and missing business context. That may not sound dramatic, but it is often where real systems succeed or collapse.

AI coverage often gets pulled into model chatter. Which one is faster, cheaper, or better at reasoning. This session moved away from that and looked at the infrastructure underneath. That shift mattered, because a lot of AI trouble begins long before a prompt reaches a model.

Companies want AI systems that can support decisions, trigger actions, and respond as even...