What happened when we let AI send a campaign without an approval gate, and what we built after

Early in building dolv, we ran an internal test without the approval workflow. AI drafted, AI sent. No human in the loop.

It did not go catastrophically wrong. But one email went to a segment it was not intended for. The content was fine. The targeting was not.

The issue was not that the AI made a bad decision in isolation. It was that there was no moment for a human to catch the targeting error before it executed at scale.

That experience defined the approval gate as a non-negotiable part of the product, not a feature. Anything that reaches a customer waits in an approvals queue by default. You can edit it before it goes. You can reject it with a note. The gate is the product.

The opt-out exists. But loosening it is a deliberate choice, not the default.