Enterprise Architecture and Sovereignty

The information system of a large Italian company almost always tells the same story. On top, recent applications, often cloud-native. Below, a core of systems built in different times, with different technologies, often from different suppliers, held together by integrations that no one fully remembers. 

The problem emerges when the whole becomes ungovernable: a CIO can no longer confidently answer simple questions like which processes depend on which systems, where critical data resides, which suppliers have become difficult to replace, how realistically possible it is to change course. 

The cost of this loss of visibility is already measurable. CIOs estimate that technical debt represents between 20% and 40% of the value of the entire technological asset, with 10-20% of the budget for new products diverted each year to address it. Less than a third of organizational transformations succeed in improving performance in a lasting way; for digital transformations, the figure drops to 16%.

88% of companies use AI, but less than 10% of processes fully integrate AI agents. If there's no systemic design upstream, there's a risk of backtracking on what's already in production.  

Many organizations have invested in training and experimentation, with agents built case by case on ad hoc assembled contexts. That value remains tied to the person who built it and lives in the session of whoever uses it, without becoming organizational capability: 79% of companies introduce AI without systematically deconstructing the workflow to automate, the factor most correlated to the real impact on business results.  

By 2027, four out of ten companies will have to revert or deactivate agents already released due to governance issues that emerged only afterward.

For us, Enterprise Architecture does not impose top-down standards: it is a practice of sharing knowledge between ICT Management, business, and suppliers. We extend the traditional boundary of systems and data also to the agents and the decisions they need to be able to make. Our Enterprise Architects are involved from the first introductory and analysis interviews, even before the project takes shape. The role now has a precise goal**:** a person who works within the client’s context from the start, to bring out the real sources of value before designing any solutions. 

1. Mapping dependencies beyond applications 

 
A dependency matrix that connects processes, data, integrations, suppliers, and skills. If this element changes, how does it impact the business? The technical debt that emerges from this map is not just a risk to flag: it becomes the criterion for deciding what to modernize first and what, instead, is worth maintaining as it is. 

2. Classifying data by strategic value and level of control 

 
We distinguish operational and critical data, clarifying ownership, access, regulatory constraints, and portability — the first step for a credible exit plan. 

3. Linking Enterprise Architecture to strategy 

 
The work converges into a Business Capability Map, built together with business managers: it sees each initiative not only for its technical benefits but for its effects on flexibility and future decision-making capacity. 

This work does not end with the initial design. We maintain observability and control throughout the application life, up to operational management, so that decisions made in run remain consistent with the architecture defined upstream. And we do it piece by piece with a journey built together with the client, coordinating IT’s priorities with the business needs as they emerge.

From descriptive governance to decision-making governance 

 
Descriptive governance merely captures the existing: catalogs systems, applications, flows. Decision-making governance is able to enable strategic choices because it makes the truly available options explicit, not just the current state of things.

The benefits affect multiple levels of the organization, from alignment between business and IT to the freedom to revise technological choices over time. Here are the main ones: