Tracking AI Agent Actions Across Organizational Boundaries
For seventy years, AI research has shown that general methods that scale with compute beat human-designed rules. Each new model generation absorbs more of the scaffolding enterprises built around the last one. Florian Douetteau of Dataiku, who began his career hand-building language systems that deep learning later replaced, asks where this lesson stops applying. His answer: at the boundaries between parties who need to trust each other. This session examines why data lineage, agent authority, and process state matter more as work passes between agents, people, and organizations, and how Dataiku makes them explicit so companies can verify what their agents did.
CEO, Dataiku