AI is everywhere….
But while everybody is talking about AI agents, we believe there is another conversation worth having: What does it actually take to integrate these agents into an enterprise landscape and what AI foundation platform is required?
When discussing agentic AI, the focus is often on what is visible: intelligent agents, their capabilities, and the business value they promise.
However, successful enterprise adoption requires much more than deploying an AI agent.
Agents need to interact with existing business applications, access relevant information, exchange data, and participate in end-to-end business processes. This introduces additional requirements across multiple dimensions, including data management, security, and integration, that go far beyond the AI agent itself.
And instead of focusing only on the tip of the iceberg, why don't we turn it around?
Turning the Iceberg Around: Integration Takes Center Stage
Turning the iceberg around means shifting our perspective.
Rather than starting with the AI agent and treating integration as a secondary consideration, we bring the integration architecture into focus from the beginning.
This raises several important questions:
- How will AI agents interact with existing SAP and non-SAP applications?
- Which integration technologies, protocols, and patterns are required to enable agentic interactions?
- How can AI agents securely discover and access enterprise data, APIs, and data products?
- How can we ensure AI agents have seamless access to accurate, trusted, and high-quality enterprise data through reusable, well-governed data products?
- How can we ensure secure agent-to-system communication, including principal propagation, while maintaining end-to-end observability across agents, integrations, and backend systems?
- What architectural requirements emerge when introducing agentic AI?
- How mature is the organization's integration landscape in supporting these scenarios?
These questions are particularly relevant because enterprise landscapes are rarely homogeneous. Organizations typically operate a combination of SAP and non-SAP applications, different integration technologies, and established integration patterns.
The challenge is not simply to connect an AI agent to an application. It is about establishing the right AI foundation – including trusted data, secure access, and integration capabilities – and ensuring that agentic AI fits seamlessly into the overall enterprise integration architecture.
From Agentic Projects to a Systematic Integration Assessment
Based on our experience with agentic projects, we have been applying the SAP Integration Solution Advisory Methodology (ISA-M) as a systematic approach to evaluating integration requirements and performing technology mapping for agent integration.
Instead of evaluating integration technologies in isolation, ISA-M provides a structured methodology for looking at the broader integration landscape.
This allows us to approach agent integration from an architectural perspective, connecting individual technical requirements with the organization's overall integration strategy.
Our objective is to move from individual agentic use cases toward a more comprehensive understanding of the technologies and capabilities required to support them.
In other words, we want to make the entire iceberg visible.
The assessment helps us identify existing capabilities, relevant technology requirements, and potential areas for improvement. An important aspect of this approach is that it is not limited to SAP agents.
Rather than immediately recommending individual technologies, we first aim to establish a shared understanding of the current integration landscape and the requirements introduced by agentic AI.
The results provide the foundation for two important deliverables.
- An overall maturity matrix
The maturity matrix helps visualize the assessment results across the evaluated levels. It creates transparency about the current state and highlights areas that may require further attention.
2. A joint integration roadmap
Based on the assessment and maturity matrix, we work together to define a roadmap that connects identified requirements with potential integration modernization activities.
This creates a structured path forward instead of approaching every agentic integration scenario as an isolated project.
From the Tip of the Iceberg to the Bigger Picture
Focusing exclusively on AI agents risks overlooking the architectural foundation needed to integrate them into real business processes.
By turning the iceberg around, we can shift the discussion from individual AI capabilities toward the broader integration requirements that enable them.
With ISA-M as a systematic methodology, our approach combines technology mapping, an AI foundation assessment, a maturity matrix, and a jointly developed roadmap.
Are you currently exploring agentic AI scenarios and wondering how they fit into your existing integration architecture? We would be happy to discuss our approach and explore how an integration assessment could support your journey.
Reach out to your SAP Customer Success Manager (CSM), your SAP Account Team, or contact the SAP Integration Modernization team directly:
integration.modernization@sap.com
Let's explore what is above and below your iceberg – together.
