Query your Postgres database from a workflow

Your production database already knows the answer to most operational questions: which trials expire this week, which orders are stuck, which accounts went quiet. The Postgres integration lets a workflow ask those questions with real SQL over a direct connection and act on the rows: post them to Slack, run them through an AI agent, email a digest, or write a result back. Bulk updates and deletes can wait for a yes from a person first, and every run records each step's rows in and rows out, so a wrong update traces back to the query and input that caused it. Describe the job in plain English and MadMax writes the parameterized query, wires the connection and runs the step against your database during the build so you see real rows before anything is scheduled.

What teams build

• An 8 am schedule runs execute_query for trials ending within three days, a loop hands each row to an AI agent that drafts a specific note from the account's usage columns, send email delivers it, and update_row stamps last_contacted_at.
• Every 30 minutes a query finds orders stuck in processing for over two hours, the list is posted to Slack, and an approval asks ops whether to auto-cancel; on approve, update_row sets them to cancelled and customers are notified.
• A signup webhook runs an AI classifier for industry and size, a query checks for an existing row by email, and an if step chooses update_row or insert_row.

How it connects

Fill in host, port, database name, username and password, and turn on Require SSL for managed providers (RDS, Neon, Render, Fly and similar need it). The host must be reachable from MadMax, and a read-only database user is a good first connection. Credentials are stored encrypted and never rendered back; changing the password is a reconnect on the same connection. Table and column names are validated and values are always bound as $1, $2, … parameters, so data from a form or a webhook never gets interpolated into the SQL string. See managing connections.

Related

• Schedule trigger
• Approval step
• Data stores
• Managing connections
• Supabase