Row creation and updates

Structured data only helps when it stays current. When an upstream event carries new information, the agent reads it, maps the fields to the matching Retable columns under your approved schema, and writes the row or updates an existing one. It confirms the write and notes anything it could not map. Records that fail validation, duplicate a key, or carry values outside your allowed ranges are held for review instead of saved. Teams keep their tables fed from other systems without someone retyping every entry by hand each day.

Field validation

A database is only as good as the values inside it. The agent reads each incoming or changed row, checks the fields against the rules your team approved, required values, formats, allowed options, and marks clean rows as verified. Rows that break a rule are set aside with the specific problem noted, and anything ambiguous is routed to an owner to decide. It never overwrites a flagged value on its own. Bad data stays out of the table so downstream reports and views run on records people can trust.

Table lookups for other agents

Retable often holds the reference data other workflows need. When another process asks, the agent reads the relevant Retable rows, applies the filter your rule defines, and returns the matching records to that workflow. It logs which rows were read for traceability. Requests that match nothing, return more rows than expected, or touch a restricted table are routed to a person rather than answered with a guess. Your tables act as a shared source of truth without exposing the whole database to every automation.