Climent Ads Assistant
How it works
Three moving parts, and the third one is the one that matters: nothing reaches your account without passing you first.
Climent Ads Assistant runs a loop: it imports one closed day from each connected account, an agent layer reads across all of them and works out what it would do, and every proposed action stops at an approval gate with an audit trail. The reading is automatic. The deciding is not.
What you get out of it
- The import is scheduled and boring on purpose: one closed day, once a day.
- The agent layer reads across accounts, not one platform at a time.
- Every action it proposes is logged, attributable and reversible.
Where this sits
That is the whole loop: it reads on a schedule, works out what it would change, and waits. Climent Ads Assistant applies it to Meta Ads, Google Ads, Google Analytics and Search Console in one pass. Apache-2.0, with a runnable demo on synthetic data.
FAQ
Where does the agent layer come from?
From Paperclip, the internal platform we built to run every agent in one place with structure and oversight. Its whole design principle is that approval gates and audit trails sit at the centre rather than being bolted on afterwards. The ads console inherits that, which is why the approval step is not an option you can switch off.
What happens if the agent is wrong?
You decline the proposal and it costs you the ten seconds it took to read it. That is the entire point of the gate. An agent that acted first would have spent the money before you had an opinion.
Can I turn the approval step off?
No, and that is deliberate. If you want an agent that acts on its own, there are tools built for that and they are honest about it. This one is built for accounts where being wrong is expensive.
Last updated: