Insights from Gartner’s 2025 IT Symposium/Xpo reveal that the central challenge for enterprises is no longer experimentation, but successfully scaling initial AI pilots into a secure, well-governed, and fully integrated component of the enterprise technology stack.
The conversation around artificial intelligence has decisively shifted from theoretical exploration to pragmatic, enterprise-wide deployment, a key theme at the recent Gartner IT Symposium/Xpo in Barcelona. While the prior year was dedicated to small-scale learning, the focus now is on scaling AI for end-to-end transformation. Successfully navigating this transition requires addressing five critical areas that extend far beyond simply deploying a large language model.
The Emergence of the AI-Ready Technology Stack
Scaling AI agents necessitates a fundamental rethinking of the core enterprise architecture, recognizing that AI will primarily function through coexistence with existing systems of record, notably Enterprise Resource Planning (ERP) platforms.
Integration and Process Redesign: A traditional, structured system of record will remain the indispensable backbone of back-office processes. Full AI transformation is therefore contingent on upgrading these underlying systems and reworking future processes for AI, rather than merely applying automation to legacy workflows. This demands a redesign of processes to leverage the flexibility of a “headless” ERP environment.
The Agent Ecosystem: Companies face complex build/buy/partner decisions regarding their future agent workforces. While leading software vendors are embedding native AI agents into their applications, a truly scalable enterprise stack requires support for multiple agents. This interoperability is achieved through Model Context Protocol (MCP) services, which ensure agentic systems of intelligence can effectively dock with and communicate across various systems of record.
Security by Design: Initial AI experiments may have prioritized speed, sometimes overlooking crucial security protocols and exposing enterprise data. As use cases scale, the future stack must be secure by design. Technology leaders must rigorously comply with organizational security frameworks and enforce strict Know Your Agent (KYA) protocols. KYA verifies that an agent authentically represents its stated person or company and is authorized to execute its intended task, addressing critical issues of identity and permission in an autonomous environment.
Data Quality: The Foundational Challenge and Opportunity
Despite the technological advancements, data remains the biggest bottleneck and, simultaneously, the greatest source of opportunity. Autonomous agents rely entirely on clean, organized, governed, and readily discoverable data to be effective and trusted. Data quality must be managed meticulously at the source.
Data Contracts: To ensure reliability, data contracts must be established to guarantee that information is complete, accurate, and regularly refreshed within the system.
Lineage and Provenance: Clear and transparent lineage and provenance provide essential confidence in the data’s origins and its journey through the enterprise, a non-negotiable requirement for regulatory compliance and trust.
Investing in Workforce Education and Change Management
The successful deployment of AI is ultimately a human challenge. Regardless of whether AI ultimately creates or eliminates jobs, demographic changes will intensely escalate the competition for the best AI talent.
Training Investment: While technology departments are keen to deploy use cases, scaling AI requires pervasive adoption and understanding by business users. Some industry experts estimate that companies may need to spend as much as twice the amount on training and education as they spend on the underlying technology itself.
Rethinking Change Management: Because technology providers and systems integrators have historically lacked strength in change management, enterprises must rethink deployment programs and intentionally build up their internal capacity for managing large-scale organizational transition.
Geopolitical Shifts and the Repatriation of Technology
A growing trend mirroring shifts in physical supply chains is the increasing wariness of relying excessively on foreign suppliers for critical processes and sensitive data storage.
Sovereign Cloud and Localization: To build resiliency against geopolitical shocks, there is a heightened interest in sovereign cloud and local solution providers. Hyperscalers are responding to this by expanding their global footprint.
Edge Processing: For AI specifically, increasingly stringent data privacy requirements and regional regulations are driving the necessity for data to be processed locally, often at the network edge. This localization will shape AI training and inferencing, potentially introducing new risks of algorithmic bias that require close monitoring.
The Elevated Role of the Chief Information Officer
Far from receding into the background as technology democratizes, the Chief Information Officer (CIO) is assuming a crucial, elevated role in the AI transformation. With a large percentage of current AI initiatives failing to achieve a break-even point or generate value, the CIO is the necessary steward for strategic clarity.
The CIO’s mandate is to set the ambition and systematically assess the right level of automation for every business domain. This assessment must differentiate between deterministic automation, agentic AI, and transformation achieved through AI copilots, all while maintaining a fundamentally human-centric perspective. With this integrated understanding, the CIO can articulate a clear, full-potential ambition for the enterprise. Crucially, they must reinforce the importance of a fail-fast culture, allowing for the rapid abandonment of unpromising ideas to concentrate resources on scaling the most valuable, transformative initiatives.
