Event sourcing doesn't require a greenfield project, micro-services, kubernetes or a second database. The Trezy team introduced it step by step on top of a Rails monolith with Sidekiq and a single Postgres database: first storing webhooks as technical events, then adding aggregates for critical invariants, then separating read/write models only where the business said it mattered. Event-driven architecture on top of that can be risky (no order guarantee!) and should be used where it makes sense.