Table row updates

NocoDB stores your data in tables whose rows behave like spreadsheet entries. A Beam agent watches a table for new or changed rows, reads the fields, and applies the rule your team approved, for example setting a status or filling a linked field once the required cells are present. It writes the result straight back to the row and notes what it changed. When a row has conflicting values, missing keys, or data that does not fit the rule, the agent leaves it untouched and routes it to a person to decide instead of guessing at the answer.

Grid and gallery views

NocoDB presents the same table through grid, gallery, and other views that filter and group rows for different teams. A Beam agent reads the view your workflow depends on, checks which rows now meet a saved condition, and updates the fields your rule covers, such as marking an entry ready or reassigning an owner. It notifies the channel your team names when a batch is done. Rows whose values look unusual against past entries, or that would move an item into a stage a human owns, are set aside for review rather than changed on their own.

Linked records across tables

NocoDB lets one table reference rows in another so related data stays joined together. A Beam agent reads a parent row, follows the link to its children, and applies your approved rule across them, for instance copying an agreed field or updating a shared status when the parent changes. It writes each update back and tells your team which links it touched. If a link is broken, points at several candidates, or the change would affect records outside the agreed set, the agent stops and hands the case to a person to sort out.