Integration Patterns

Integration Patterns, most famously catalogued as the Enterprise Integration Patterns (EIP) by Gregor Hohpe and Bobby Woolf, are a shared vocabulary for connecting software systems that were never designed to talk to each other. Rather than inventing a new approach for every point-to-point connection, teams reach for named, well-understood patterns — message channels, routers, translators, and aggregators — that describe how data and events move between systems reliably.

• Messaging patterns - Channels, message routers, and dead-letter queues that describe how discrete messages flow, retry, and fail between systems.
• Data transformation - Translators and canonical data models that reconcile the different formats each system speaks so nothing is lost in the handoff.
• Service orchestration - Process managers and content-based routers that coordinate multi-step work across several services into a coherent business flow.
• A shared vocabulary - Named patterns that let architects communicate an integration design without redrawing the same plumbing from scratch every time.

In real API operations these patterns are the connective tissue underneath API gateways, event buses, and webhook fabrics — the same building blocks I see providers reach for when they move from simple request/response toward event-driven APIs and workflow orchestration. As the agentic turn pushes more machine-to-machine traffic through APIs, the old integration patterns are resurfacing as the design language for how agents chain calls, transform payloads, and route work across many providers.