Text-to-SQL promised self-serve analytics but failed to deliver trust. Despite impressive demos, these tools consistently break in production—hallucinating business logic, misinterpreting context, and producing results teams can't rely on. The fundamental problem is that these systems lack semantic understanding of organizational data.

This talk explores what comes next: AI analytics built on semantic foundations that go far beyond simple query translation.

You'll see real examples of semantic architectures, understand how agentic systems differ from traditional tools, and learn to identify the patterns that separate reliable AI analytics from expensive experiments.