Integration in Three Layers: A Timeless Model
To understand this evolution, it’s worth recalling that modern integration is structured around three distinct layers:
- Technical connectivity: the “plumbing” that enables systems to communicate
- Domain services: standardized and reusable business components, foundations of composable architectures
- Composition and orchestration: the intelligent chaining of processes
Current marketing massively focuses on this third layer, particularly with the rise of agentic AI.
SOA Revisited: Principles Still Relevant Today
This layered approach is not new. It draws directly from SOA (Service-Oriented Architecture) principles which, despite technological evolution, remain relevant. SOA addresses a constant need for information system urbanization, with objectives of modular, scalable, and resilient architectures.
What has changed is how to achieve these objectives: architecture decentralization, increased velocity, finer granularity, and “as code” governance. But the fundamentals remain.
Integration vs Automation: A Crucial Distinction
It’s essential to distinguish these two concepts:
- Integration aims to connect and harmonize heterogeneous systems
- Automation aims to execute processes with minimal human intervention
It’s equally important to emphasize that integration is a powerful enabler for automation, particularly for AI. Without a mature integration layer that guarantees unified data access and consistency, no intelligent automation is possible. Agentic AI entirely relies on this capability to access the right information at the right time.
The Post-Covid Context: New Constraints, New Game
Regulatory explosion (GDPR, NIS2, AI Act), growing security threats, digital sovereignty issues, and cost control have radically changed the context. Business teams can no longer focus solely on value creation; they must integrate these constraints into their processes.
This evolution creates a major challenge: while the number of applications and data volume explode, and tools have adapted to business agility, IT must now reintroduce control without breaking the value creation dynamic.
Platform Evolution: Two Models Behind iPaaS
Facing these new challenges, two distinct models emerge behind the generic “iPaaS” label:
The automation model focuses on business automation and targets teams seeking autonomy through graphical interfaces and rapid deployment. However, it presents structural weaknesses: limited governance, difficulties implementing reusable domain services, and “point-to-point” logic.
The hybrid integration model represents the evolution of traditional integration platforms toward hybrid and multipattern solutions. It implements platform engineering principles to recentralize governance while maintaining agility benefits. This approach enables central governance (security policies, architecture standards, service catalog) while distributing execution (self-service, pre-approved templates, automated deployments within guardrails).
In reality, many solutions attempt to reconcile these two models with varying degrees of success, gradually evolving from one to the other based on their maturity and market positioning.
Conclusion
Before blindly following analyst recommendations focused on automation, companies must ask the right questions: are their integration foundations solid? Is their governance adapted to new challenges? Do they have the necessary architectural maturity?
Don’t mistake objectives: automation is a means serving operational efficiency, but it’s only possible on solid integration foundations. Modern platforms must certainly meet business agility needs, but within a controlled architectural framework that guarantees governance, security, and compliance.