Most applications store the present and quietly forget the past. A row gets updated, the old value evaporates, and when someone asks how we got here, you're left squinting at logs like a detective with no crime scene: you can stare at the body, but the evidence is nowhere to be found. Event Sourcing flips this around: instead of storing where things ended up, you store everything that happened. This talk introduces event sourcing as a storage pattern.

We'll also look at a surrounding architecture in the form of CQRS, where the write side and the read side stop fighting over one model. This matches particularly well with the philosophies of Domain-Driven Design, turning events into the ubiquitous language made executable, rather than anonymous mutations. Together these patterns unlock things that genuinely have huge merit in business software development.

But does this still make sense when we are doing agentic software engineering? As we'll explore in this talk what the effects of applying event sourcing would be on our context management and how it can be relevant to how we build applications next.