CEO agent triages a deliverability incident

This hands a live deliverability alert to a CEO agent that reasons over the full picture rather than a single threshold. It gathers bounce rate, complaint rate, recent send velocity, and domain history, then chooses the proportionate response: throttle the send rate, pause one segment, or halt the whole campaign.

Use it when blunt threshold rules cause too many false pauses and you want judgment in the loop. Best for mature programs where over-pausing costs revenue but under-reacting costs reputation.

1. 1A deliverability alert posts to the webhook with the campaign and triggering metric.
2. 2The agent queries Postgres for that campaign's recent bounce, complaint, and volume history.
3. 3It weighs severity and trend, then decides between throttle, partial pause, or full halt.
4. 4It executes the chosen action against the ESP API.
5. 5It writes a structured incident brief to Notion explaining what it saw, what it did, and what a human should verify, then notifies Slack with the link.