Natural language query (NLQ) systems combine language models with the semantic layer.

When a user asks, ‘show revenue by region,’ the system parses the intent, resolves entities like ‘revenue’ and ‘region’ to actual tables or metrics, and constructs the corresponding SQL.

The semantic layer provides metadata—definitions, joins, filters—that ensures the generated query is accurate and secure.

The system then executes the SQL against your data warehouse and returns a visualization or narrative explanation.

High‑quality NLQ requires training on domain‑specific language and robust guardrails to prevent unsafe queries.

MageMetrics builds a ‘compounding intent‑to‑SQL’ flywheel, learning from each query to improve future translations.