Semantic Model
A data abstraction layer that centrally defines business metrics and entities in one place, including the precise tables, joins and filter conditions behind each metric, so that AI agents can query metrics by name rather than writing SQL directly, ensuring consistency and accuracy of business definitions.
Timeline (last 90 days)
Semantic models are suited to recurring business metrics, not to one-off questions about a single document, nor to scenarios that connect facts across entities.
There are currently five approaches to structured data for AI Agents that are actually used in production environments: semantic models, ontologies and knowledge graphs, retrieval-augmented generation (RAG), plain Markdown files with grep search, and open knowledge formats.
In dbt Labs' 2026 benchmark, an Agent querying a semantic model reached 98.2%–100% accuracy on a set of business questions, while the same model writing raw text-to-SQL against the full schema achieved 84.1%–90.0% accuracy.
0sql's semantic model defines four types of elements: measures, dimensions, join relationships, and row-level security filters.
0sql's semantic model is a business semantic contract described in YAML or JSON, usually stored under version control in a Git repository.
Semantic models answer aggregation-type questions (such as rankings and counts), while ontologies answer relationship-type questions (such as multi-hop relationship queries)
Semantic models are primarily designed for human analysts, assuming that users already have the background knowledge to understand field naming and hierarchical relationships.
Semantic models are more like relational mapping tables (field A corresponds to metric B), whereas ontologies are more like knowledge graphs that explicitly encode inheritance relationships, mutual exclusion constraints, and temporal validity between concepts.
All Facts (8)
Semantic models are suited to recurring business metrics, not to one-off questions about a single document, nor to scenarios that connect facts across entities.
50%UnverifiedThere are currently five approaches to structured data for AI Agents that are actually used in production environments: semantic models, ontologies and knowledge graphs, retrieval-augmented generation (RAG), plain Markdown files with grep search, and open knowledge formats.
50%UnverifiedIn dbt Labs' 2026 benchmark, an Agent querying a semantic model reached 98.2%–100% accuracy on a set of business questions, while the same model writing raw text-to-SQL against the full schema achieved 84.1%–90.0% accuracy.
50%Unverified0sql's semantic model defines four types of elements: measures, dimensions, join relationships, and row-level security filters.
50%Unverified0sql's semantic model is a business semantic contract described in YAML or JSON, usually stored under version control in a Git repository.
50%UnverifiedSemantic models answer aggregation-type questions (such as rankings and counts), while ontologies answer relationship-type questions (such as multi-hop relationship queries)
50%UnverifiedSemantic models are primarily designed for human analysts, assuming that users already have the background knowledge to understand field naming and hierarchical relationships.
50%UnverifiedSemantic models are more like relational mapping tables (field A corresponds to metric B), whereas ontologies are more like knowledge graphs that explicitly encode inheritance relationships, mutual exclusion constraints, and temporal validity between concepts.
50%