Approved-rule row updates

When an upstream event arrives, such as a closed order or a corrected address, a Beam agent opens the relevant Azure SQL table, reads the current row, and applies the customer-approved rule for what may change. It writes the new value, records which rule fired, and leaves an audit trail your team can inspect later. Straightforward edits go through without a wait. If two rules conflict, a required field is missing, or the change touches a protected column, the agent stops and routes the row to a person for a decision.

Read-only lookups

Agents run read-only queries against Azure SQL when another workflow needs a fact, such as a customer's plan or an invoice balance. The agent reads the result, checks it against the approved rule for what counts as valid, and passes clean values downstream to update a ticket, an email, or a second system. Nothing is written during a lookup, so the risk stays low. When a query returns no match, several matches, or a value outside the expected range, the agent flags the case and asks a person how to proceed rather than guessing.

Scheduled integrity checks

On a schedule you set, an agent scans chosen Azure SQL tables for rows that break an approved rule, such as a missing region code or a status that has sat unchanged too long. It reads each flagged row, corrects the ones the rule clearly covers, and compiles the rest into a short report. Small, unambiguous fixes are applied directly and noted. Anything that looks like a data-quality problem beyond the rule, or that would touch many rows at once, is held back and sent to a person before any change lands.