Why dbt Labs thinks AI agents will finally make the semantic layer stick

dbt Labs staff product manager Zach Mandell argues that AI agents could revive adoption of semantic layers by making governed data accessible through conversational analytics, though trust remains a major challenge because errors compound across multi-step agent workflows. He says the first wave of semantic layers, focused on centrally defining metrics as code for BI tools, stalled due to difficult integrations, limited incremental value, and the unfamiliarity of data modeling concepts for many analytics engineers. For a second wave to succeed, Mandell proposes native “semantic SQL” access, deeper integration of semantics into data transformations and pre-aggregation, a deterministic semantic layer within a broader context layer for structured and unstructured data, and tightly scoped permissions with observability into agent data access. He also noted dbt’s involvement in the Open Semantic Interchange initiative, acknowledged unresolved interoperability challenges among semantic-layer specifications, and emphasized that semantic models should be human-curated rather than generated solely by AI.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.