Eric Siu explains AI loops, a pattern for automating work by measuring a target state, observing the current state, and compounding improvements over time. He walks through a conversion rate optimization loop aiming for a 30% lift in four weeks, and introduces a "loops creator" skill that enables autonomous learning.
Everyone's talking about AI loops right now, but nobody has really explained it well, especially as it relates to business. So the way a loop works is that it is able to measure and improve on the work that it's doing. So it needs to see a target state, it needs to see the observing current state, and it needs to be able to see the gap over time so it can continually improve on the gap. And then a loop compounds over time. It's a lot different than what you have with an agent. And so what we did was we built a loop for Single Brain, which is one of my websites. And SingleBrain, the loop, it's a CRO loop, meaning it optimizes the conversion rates. So every week it'll look at the conversion rate. It'll say, let's say it's 1.80%. Then I put in a loop saying, hey, I want to lift this by 30% in the next 4 weeks or so. So it's very verifiable and you can see it has a way to learn over time. And our loops creator skill has the ability to create the loop. So loop can ultimately compound and learn on itself versus a standard prompt really can't. And a slash goal typically can't. So that is the difference between a loop versus the other things that people are talking about. It's extremely powerful. By the way, if you want our AI work operating system as it relates to all this loop stuff, just comment Hermes and we'll send it over to you.
