An InfoQ presentation by Aditya Kumarakrishnan argues that AI teams need to move beyond the hype cycle toward stronger foundations for agent systems. The talk covers modular frameworks, process science, and event-sourced artifacts.
The message is relevant as companies try to turn agent demos into durable workflows. Agents operating across departments need memory, process structure, and reliable state rather than one-off prompt chains.
The presentation frames agent engineering as an infrastructure and workflow problem, not just a model-selection problem.