Ask us the hard questions first.

Q1— So an AI reads my posts. Why would I trust an AI to catch what an AI got wrong?

Because the checker isn’t allowed to freelance. Every finding must include a verbatim quote from your post. If it can’t point to your actual words, the finding is thrown away automatically. And required disclaimers are verified by plain deterministic checks — string matching, not judgment — that run even if the AI review fails entirely.

Q2— What rules does it actually know?

Today: a Medicare Advantage rule pack built around CMS marketing requirements — prohibited claims, required disclaimers, the phrasing traps. Your brand carries its own compliance profile, so the review runs with your context, not a generic one.

Q3— And when it finds something?

The post is blocked from approval. Not flagged-and-forgotten — blocked. You read the finding, you fix the text or you acknowledge the finding against your attestation text, and only then does the approve button unlock. The acknowledgment is stored with your identity, a timestamp, and a fingerprint of the exact content you reviewed. Edit the post afterward and the review runs again — the record always matches the words.

Q4— Who approved what — can I actually produce that?

Yes. Every post keeps its full version history: the AI’s original, every edit, every regeneration, who approved it, who acknowledged which findings, when. If anyone ever asks “what did you review and when,” the answer is a record, not a recollection.

Q5— Does this make me compliant?

No. It assists supervision; it does not certify or guarantee regulatory compliance. You are the licensed professional; every post still crosses your desk. What changes is that your desk got a reviewer that never gets tired, never skims, and never forgets to write it down.

Q6— What can it post without me?

Nothing. Publishing is off until you enable it, and even then only approved posts move. There is no autopilot mode to accidentally leave on.