Event Sourcing is a software pattern: we record application state as a sequence of events throughout the time. Events could be then be projected to create application state at any moment in time. We could even derive new facts from the events.
This pattern has a few benefits.
I’ve learned about Event Sourcing from Greg Young, then used in multiple projects (including high-load e-commerce), taught courses and wrote a chapter on Event Sourcing for the “Implementing Domain-Driven Design” book.
In pre-AI Coding era I mostly reached out to event sourcing in complex business software that required precise tests, rapid development with special scalability and availability needs. In projects like that the hassle of setting up event sourcing in a project was worth the benefit.
Things changed around the beginning of 2026, when agents became powerful and smart enough to take over the mundane parts of software engineering in this approach, while putting a human into the seat of a product owner and domain expert.
Things that worked well in pre-Codex/Claude era with a bit of coding effort and due diligence, work amazingly well in modern days. So let’s revisit Event Sourcing, while accounting for the modern AI Native project nuances.