AgentNLQ: A General-Purpose Agent for Natural Language to SQL

In three linesAgentNLQ, a multi-agent method, achieves 78.1% semantic accuracy on the BIRD benchmark for natural language to SQL conversion. The system uses an optimized orchestrator to plan, reflect, and self-correct queries, enriches schema with context-aware metadata, and incorporates user-provided business rules.

Summary generated by Claude — human-verified