The BEAM makes heavily evented systems incredibly easy to implement. However, it's just as easy for these systems to devolve into asynchronous spaghetti. LLM's are not making that any better. How can you maintain a clear understanding of cause and effect in a system that relies heavily on pub/sub?

In this talk, we'll explore techniques to understand the action-at-a-distance that we all love. We'll talk about defining clean APIs for publishing and subscribing to events. We'll explore techniques to enforce as much clarity as possible at compile-time, letting us know as soon as possible that our code is invalid.