Rethinking the Semantic Layer- Part III: Is There Still Room for Semantic Tools?

​Semantic layers have long helped teams turn business concepts into consistent queries and metrics. Agents change how those definitions are consumed: they can generate SQL, write code, and reason through complex scenarios, but they still need to understand what the data means. We are back with a new fireside chat on a central data question: where should business meaning live, and how should it be shared between humans and agents?

The launch of Bauplan SDK 0.3 gives us a concrete starting point. Typed contracts now span SQL and Python pipelines, and semantic definitions are stored in Iceberg metadata and versioned with the data. As more context becomes part of the code agents already work with, what responsibilities remain for a dedicated semantic layer? Are new standards like Ossie already being leapfrogged by current models?

Join us on November 17 for a fireside chat with:

• ​Julian Hyde, author of Apache Calcite and the Mondrian OLAP engine

• ​Matthew Mullins, CTO at Coginiti

• ​Moderated by Jacopo Tagliabue, CTO and co-founder of Bauplan.

​We’ll explore the relationship between who or what produces the tables and who or what consumes the data, and how that handshake should be built and maintained. In other words, is there still room for dedicated semantic tools now that coding agents can consume unstructured data and code directly?