AI agents have the potential to transform both organizational structures and operational processes. They enable companies to increase quality and speed to a level that cannot be achieved with traditional ways of working.
According to an IBM study, 93 percent of surveyed German executives believe that agent-based AI will deliver measurable returns within the next two years. This expectation highlights the strategic potential companies see in agent-based AI solutions. At the same time, practical experience shows that many AI agents remain isolated pilot projects or are limited to individual use cases. The reason is rarely the AI technology itself, but rather the question of which data basis these agents operate on and how they are integrated into existing processes.
In principle, two layers can be distinguished: the action or process layer (Action Layer) and the data context layer (Data Layer). This article focuses on the latter.
AI Agents Require More Than Algorithms
AI agents place different demands on the use of data than traditional analytics or reporting solutions. While reporting systems are typically based on clearly defined questions, data models, and KPIs, AI agents operate in a task- and context-driven way.
Although they also access structured data sources, they must decide depending on the situation which information is relevant and how it should be connected and interpreted.
To do so, AI agents need not only accurate data, but also a consistent, up-to-date, and domain-specific data context. The more heterogeneous the data landscape and the more information is distributed across different systems, the more important it becomes to establish a shared data foundation that agents can access in a controlled and traceable manner.
This makes one thing clear: AI agents are not an isolated AI topic. Whether they can be deployed effectively depends on the quality of the underlying data foundation and on how data is organized, maintained, and used within the enterprise.
Data Platforms as the Basis for Scalable Agents
Modern data platforms create the essential prerequisites for ensuring that AI agents not only work in individual scenarios, but can be sustainably integrated into processes and scaled beyond single business units.
A central data platform consolidates data from multiple sources and establishes core security capabilities that support the definition and enforcement of fine-grained access rights. Responsibilities and governance rules should be defined from the outset and closely embedded into the platform. For AI agents, this means they can access curated data, operate within defined rules, and evolve in a controlled way.
Modern data platforms based on data lake approaches can also serve as the link between structured and unstructured data. Especially with unstructured data, an additional transformation step is often required before it can be effectively used.
Third-Party System Connectivity as a Key Factor
In practice, another central challenge quickly becomes apparent: AI agents rarely work exclusively with data from a single system environment. Business-critical information is often stored in third-party systems such as ERP, ITSM, or specialized business applications. Selective interfaces or individual integrations reach their limits quickly. They are costly to maintain, difficult to scale, and increase the complexity of governance and security.
A central data platform therefore plays an intermediary role: it brings together data from different source systems, harmonizes it, and provides it in a unified context.
On this basis, AI agents can act across systems in a scalable, traceable, and secure way and become usable enterprise-wide.
Platforms Like Microsoft Fabric as an Enabler
Modern data platforms such as Microsoft Fabric address exactly this need. They bring data and analytics workloads together in a shared environment and create a unified data foundation through central data storage such as OneLake. At the same time, they integrate governance, security, and identity mechanisms that are indispensable for the productive deployment of AI agents.
How this approach proves itself in practice is illustrated by a project from the collaboration between Campana & Schott and Transgourmet. The goal was to build a central data platform that consolidates data from different upstream systems, standardizes it, and makes it usable for analytics and reporting purposes.
The focus was not only on technical consolidation, but on creating a resilient data foundation that enables the future use of AI applications. Today, the platform provides the basis for delivering data across systems, centrally implementing governance and security requirements, and meaningfully integrating future AI agents into existing processes.
The decisive factor is not an individual feature, but the interaction of all components. When data integration, analytics, governance, and AI capabilities converge on one platform, a consistent context emerges. AI agents can access this data without requiring new interfaces, special solutions, or separate security concepts for every scenario.
Organization and Responsibility Come Into Focus
With the deployment of AI agents, not only technical architectures change, but also organizational questions.
Who is accountable for the outcomes produced by an agent? Which decisions may it prepare or automate? And how can traceability of results be ensured?
These questions cannot be answered through technology alone. They require clear responsibilities, aligned processes, and a shared understanding of how data and AI should be applied in the enterprise. Governance thus evolves from an abstract rule set into an operational steering instrument.
Companies that want to successfully deploy AI agents should therefore think about platform, organization, and data culture together.
Conclusion: AI Agents Build on a Stable Data Foundation
Without a centrally orchestrated data platform, agents risk remaining isolated and difficult to scale. When data is consistent, contextualized, and securely available, AI agents can unlock their full potential. Data platforms therefore provide the foundation on which agents can be meaningfully built.
Companies that recognize this connection early do not only create the basis for deploying AI agents. They strengthen their overall data and AI strategy and lay the groundwork for the sustainable evolution of data-driven use cases.
Against this backdrop, Campana & Schott supports enterprises in shaping this path in a structured way — from developing a robust data and AI strategy to building central data platforms and anchoring adoption within the organization.
The goal is to treat AI applications not as isolated experiments, but to integrate them sustainably into processes and ways of working.
Are you planning to deploy AI agents or strategically evolve your data platform?
