Why a semantic layer is the difference between a 96% and an 80% accurate AI analyst

Alex Noonan of dbt Labs argued at dbt Summit 2026 that while large language models can understand natural-language questions, reliable AI analytics depends on grounding business terminology in well-defined data context. dbt Labs benchmarking found that models using a semantic layer achieved 96–100% accuracy, compared with 80–91% when generating raw SQL against database schemas, a gap Noonan described as potentially producing confidently incorrect business answers. He identified semantic layers, ontology-style entity models, and dynamic agent context layers as emerging approaches, while emphasizing that durable systems also require data testing, freshness and anomaly monitoring, machine-readable documentation, and consistent modeling. Noonan distinguished stable, reusable “just in time” context from search-based “just in case” context for unstructured information, recommended public chat channels to make AI mistakes visible and correctable, and warned that model-based evaluation can reproduce the same blind spots as the systems being judged.

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.