Segment on your own columns.

The standard architecture is a pipeline: warehouse to sync tool to marketing platform, a copy of your users maintained in someone else's schema, always a little stale. GenieOS skips it. Connect a production database from the audience surface and a segment is a predicate evaluated against your live columns. Nothing is exported, and nothing waits for a sync window.

The same connection powers measurement, and it goes further than segmentation. Bind a campaign to a column in your own database, users.activated_at say, and movement in that column is attributed back to the campaign. Success is your number in your schema, not an open rate in ours, and it is the number the loop reads before it derives the next campaign.

Contacts carry their history in a sources column: pasted in, arrived through a send, added by hand. It reads as a small detail until the day you need it, at which point it is the audit trail: why is this person on this list, and who put them there.

It is the same idea as derived, one level down. Artefacts know the campaign they came from; contacts know the route they arrived by. The whole system runs on lineage you can query, because automation you can read is the only kind worth trusting at scale.