AI Natural Language DB Queries with Dual Agents + PostgreSQL (n8n)

Turn Natural Language into Safe PostgreSQL Results with Dual AI Agents (n8n)

This n8n workflow acts as an AI database assistant: users ask questions in plain English, and it generates and executes PostgreSQL queries—while using a dual-agent architecture to keep responses safe and limited so your system doesn’t crash.

What this workflow does

• User request: A user asks something like “Show me top 10 popular products.”
• Main AI Agent: Interprets the request and applies built-in safety/query limits.
• SQL Sub-Agent: Converts the intent into a precise PostgreSQL query.
• Database execution: n8n runs the query against your PostgreSQL database and returns formatted results.
• Safe operation: Templates include important LIMIT clauses that you should keep intact during setup.

Use cases

• SaaS developers building a database search feature where users query data like they’re chatting with an assistant.
• Database administrators enabling controlled AI access to PostgreSQL without exposing risky unrestricted queries.
• Business teams who want user-friendly, ChatGPT-like database interaction for reporting and exploration.

Technical details

• AI database assistant with dual agents (Main Agent + SQL Sub-Agent).
• PostgreSQL integration to execute generated SQL and return limited results.
• n8n nodes mentioned in the workflow: set, merge, switch, postgres, telegram, sticky note.

Setup highlights

• Ensure PostgreSQL is reachable from n8n and provide connection credentials.
• Customize templates by replacing [YOUR_TABLE_NAME] and [YOUR_FIELDS].
• Copy the Main Agent system message to the primary AI node and the Sub-Agent system message to the SQL generator node.
• Keep all LIMIT clauses intact to preserve safe query behavior.