Corné Potgieter

“Human in the loop” sounds like a guardrail, but often acts as a checkbox. As AI systems scale, meaningful oversight becomes impossible and decisions degrade into superficial approval. This talk examines, through examples and anecdotes, why this approach doesn’t hold and how embedding control into systems through contracts and constraints offers a more robust path.

Data Mesh promises a bold shift—domain-driven ownership, decentralized architecture, and data-as-a-product thinking. But between the whitepapers and the whiteboards lies a messier, more nuanced story. In this talk, we’ll explore the practical realities faced by early adopters who’ve ventured beyond the theory. You'll hear lessons learned the hard way—from mismatched organizational structures to unexpected governance challenges—along with honest reflections on what worked, what didn’t, and why. We’ll contrast the “purist” vision with the pragmatist’s playbook, comparing theoretical principles to real-world adaptations across multiple organizations. Whether you’re just beginning your Data Mesh journey or already deep in the trenches, these field-tested insights will help you navigate the mesh with eyes wide open.