Oliver Laslett

In this talk, we’ll share a recipe for building a data stack that improves itself every time someone uses it, without waiting for the data team to pick up another ticket. It starts with the signals your stack already produces: what people ask, which analytics they use and where things fail. We'll show data leaders how to turn questions, usage and failures into a continuous improvement loop. That means directing people to trusted work before creating more of it, using repeated questions to identify missing metrics, repairing broken analytics automatically and removing content nobody uses. We'll also show where autonomy should stop. Agents can own work that is repeatable, observable and reversible but data teams must remain responsible for business definitions, high-consequence changes and decisions where the company has not agreed on the right answer. Attendees will leave with a framework for dividing work between agents and people, the signals and guardrails needed to run agents safely, and a plan for moving their team away from repetitive maintenance towards designing a data stack that improves itself.

Discover how Workday built an AI-powered self-service analytics platform that doesn't compromise on governance using Lightdash. They'll discuss concrete strategies for transforming your data organization from a bottleneck into a competitive advantage, backed by real implementation insights.