Claude reads your campaigns and suggests; you decide.
Every suggestion comes with its evidence from your campaigns and the quality source, and "Approve" logs your decision and opens Ads Manager for you, so nothing changes without your hand.
Three steps happen in front of you.
- 1A suggestion with its evidence
- 2Approve or ignore
- 3The decision is recorded, and Meta is in your hands

What's in Claude's suggestions with your approval
You click “Ask Claude” when you want, and it does not run in the background.
Suggests with its evidence: what it saw in your campaigns, your past decisions, the quality source and your accounts.
You click “Approve” or ignore. “Approve” records your decision under your name and opens Ads Manager for you.
Nothing changes in Meta from this screen. Execution stays in your hands.
In "Decisions" you compare the morning's suggestion with what your team actually did.
For agencies: incoming requests from customers, and read-only report links with a password and a duration.
What is asked about Claude's suggestions with your approval
Does Claude change my budgets or pause my ads?
No. It only suggests. Even "Approve" logs the decision and doesn't carry it out on Meta.
How do I know why it suggested that?
Every suggestion comes with its evidence from your campaigns, your decisions and the quality source, and whatever doesn't match your data is dropped.
What do I need to run it?
Your own Anthropic key, and an acknowledgment of AI use in settings.
Capabilities that work with it
See Claude's suggestions, with your approval, in a live demo.
A live demo on demo data, and we tell you honestly whether it suits your business.