Learning loop
The feature underneath every other feature: what happened gets written down as a reason, and the reason gets read before anything happens again.
What it does
Converts raw results into a learning — not "this got 4% engagement" but "this underperformed because the angle didn't match what the segment's reviews show they actually care about" — and makes that learning available to the next hypothesis.
How it works
This is the closing link in the autonomous loop: Hypothesis → Ticket(s) → Learning(s) → next Hypothesis. Nothing here is optional or occasional — every cycle writes a learning, whether the result was good or bad.
Why it compounds
This is the entire difference between a tool that produces content and a system that gets better at producing it. Every other feature feeds this one; this one feeds every other feature back.