What does it take to build an analytics function from zero in 2026 — when AI can answer business questions, build dashboards, and refine its own data models?

Like many fast-growing organizations, the Stellar Development Foundation began without a dedicated analytics function. Data lived in a self-hosted Metabase with no governance, and every answer was written by hand in SQL — even though the appetite for data was clearly there. This session walks through building that function from the ground up: how to create a culture of data curiosity, design a semantic layer that makes AI-driven self-service genuinely reliable, and set up context engineering and guardrails so AI doesn't hallucinate its way through your metrics.

It also gets into what AI compression of time-to-insight really looks like in practice — combining AI tools and MCPs to turn week-long dashboard builds into same-day delivery, and a semantic layer that monitors and self-builds as new data models land.

The session closes with the question every analytics leader is now facing: what does hiring and structuring a team look like when AI handles this much of the work? What should you expect from an analyst in 2026, and where does human judgment still matter most?

Attendees will leave with a practical framework for building AI-native analytics workflows, an honest picture of where AI saves real time versus where it creates new problems, and a point of view on how analytics teams need to evolve.

AI in Practice