In my recent article I tried to show how we can implement agent loops that not only implement the requested feature, but also have a meta-overview of the process and learn from failures, AND have a meta-meta-overview that learns about the learning process itself.
Fun read, thanks!
😊 thanks for feedback 🙏