Event-Driven Integration
Systems integrating by publishing and consuming events rather than by calling each other's APIs.
The advantage over API-based integration is that the producer does not need to know its consumers, so adding a consumer is a zero-change operation on the producing side — which is exactly the coupling that makes API-integrated estates hard to evolve.
The discipline that decides whether it works is event design. Publish domain facts with stable names and versioned schemas, not database row deltas and not disguised commands. Include enough data that common consumers need not call back for more (a thin event that triggers a stampede of lookups is worse than the API call it replaced), but not so much that the event becomes a coupled copy of the producer's internal model.
Operational obligations that come with it: schema registry with compatibility enforcement, dead letter queues with alarms, idempotent consumers, and a replay mechanism for when a consumer has a bug and needs to reprocess history.