Two inputs start the flow: the source content and the list of target markets. A Google Sheets node holds the per-market rule set including currency, date format, formality level, phrasing to avoid, and any regulated claim that cannot cross that border. This keeps the rules somewhere a marketer can edit without opening the flow.

A Loop node processes the markets. Inside it, a Generate Object node returns two items per market: the adapted draft and an array of changes from the source with a reason for each.

Returning the change list as a structure instead of prose speeds up review. The reviewer reads eight decisions instead of eight hundred words.

A Notion node writes one page per market so drafts do not overwrite each other and a native reviewer can comment in place. A Slack node from the Human in the Loop category notifies the reviewer assigned to that market.

An If node checks the rule set for a "do not localise" flag before generating. Content that should not cross a border produces a notification instead of a draft.

Model choice per market is worth setting deliberately. Providers differ substantially in language coverage. This is one of the few flows where the right answer is not the same model everywhere.