The next phase of AI adoption is already underway. Organizations are moving beyond chatbots and pilot projects and putting AI to work in ways that deliver real business value. By reducing the time employees spend on repetitive tasks, these agents help teams move faster and focus on higher-value work.
But as organizations begin rolling out AI agents across the business, business leaders need to think about infrastructure that will allow them to quickly scale without limits and accelerate time to value. We often see requirements change within an individual budget cycle. Weaknesses in traditional data environments and manual processes quickly become apparent:
Fragmented data foundations: AI agents are only as effective as the data they can access. When it is scattered across systems, not up to date or difficult to interpret, organizations end up relying on manual workarounds and validation checks. This can slow adoption and limit the value AI can deliver.
Outdated workflows: AI creates an opportunity to rethink how work gets done. Organizations that achieve the best results use AI to redesign workflows around speed, efficiency and better decision-making rather than layering it onto legacy processes. At GTT, we see the greatest enterprise success when we focus on getting the data right and then designing new workflows before introducing autonomous agents.
Build for flexibility
As organizations increase their AI usage, complexity also grows quickly. At GTT, we’re seeing in the world of AI tooling that requirements for new infrastructure are growing all the time. It’s often difficult for business and IT leaders to anticipate where they’ll need to invest and, just as importantly, understand how those investments will create tangible results for the business.
Typically IT leaders are procuring network and security and cloud services on a three-to-five-year timeframe. Requirements for AI infrastructure and AI applications are changing much faster, often within a one-to-two-year timeframe.
Adding to the complexity, entirely new applications and use cases will emerge over time, creating opportunities and requirements that organizations can’t foresee today.
That’s why organizations need agile infrastructure that can respond as business use cases change, integrate new technologies and support new AI capabilities.
Today’s AI environments are interconnected. AI agents need access to enterprise data and interact with core business applications. They often work together to complete tasks. All of this places greater demand on the network while creating new security considerations.
Organizations need to ensure that their agents have an identity so that their existing identity-based security tools can be used to enforce access to data and applications in the same way for their agents as they have done previously for their human workers.
At the same time, because AI models advance every few months, locking an organization into a single vendor or inflexible technology design can make it difficult to keep pace when new capabilities and evolving business requirements arise. What works today may not be the best option tomorrow.
With a modular architecture, organizations can add, replace or upgrade individual components as new technologies emerge, without having to redesign, rearchitect and redeploy. It provides the agility to keep pace with innovation while adapting to changing business needs.
Many organizations, including GTT, are adopting an AI factory approach to provide a strong operational foundation for that modular architecture. Rather than approaching AI as a series of disconnected, ad-hoc projects, this approach helps organizations scale successful initiatives more efficiently and create greater value from their data over time.
Deloitte’s Enterprise AI Infrastructure Survey: A 2028 Outlook revealed that more than 70% of the enterprise survey respondents expect to operate AI factories at scale by 2028.
GTT’s AI factory
At GTT, our AI journey started by organizing and contextualizing our internal data so it could be used effectively. From there, we built our own AI factory and GPU infrastructure, powered by Dell PowerEdge servers with the NVIDIA accelerated computing platform. To support customers with data sovereignty requirements, initially our AI infrastructure spans four geographically distributed locations. We’ve deployed more than 50 AI agents across the business to help automate processes and improve productivity.
One example is our finance cash application process. As a global organization, we receive thousands of payments every day in different currencies and formats. Historically, matching those payments to invoices was a labor-intensive task that required significant time from our finance team.
Today, an AI agent handles much of that work automatically. It reviews remittance data from multiple payment channels, matches payments to invoices, accounts for currency differences and learns from historical collections of data and business rules. What once took up to two weeks can now be completed in minutes, allowing our finance team to spend less time on manual reconciliation and more time focused on higher-value work.
This is just one example of how AI agents can create meaningful impact when they’re built on the right data foundation and integrated into well-defined processes.
For a deeper look at how we’re using AI across GTT, check out this interview with Tom Major, GTT’s Senior Vice President of Product Management and Technology, with technology analyst and influencer Evan Kirstel.
The path forward
As exciting as AI is, the real work happens behind the scenes. Organizations that focus on building strong data foundations, simplifying how work gets done and creating flexibility for what’s next will be best positioned to capture the full value of AI.
