Global Business Services (GBS) operations are evolving from cost centers to strategic hubs, and agentic AI could be the next catalyst for that shift. But the gap between promise and practical deployment is wider than many realize.

While 2025 was billed as the year agentic AI—AI capable of autonomous, goal-driven action—would take off, enterprises have struggled to move beyond pilot projects. Even tech giants like Google Cloud and development platforms such as Replit have faced hurdles in reliably deploying AI agents at scale. The same challenges that delayed generative AI adoption now apply: missing foundational infrastructure, fragmented workflows, and a lack of standardized data governance.

A survey from early 2025 revealed that 65% of GBS organizations had yet to complete a single generative AI project. Agentic AI, still in its infancy, has yet to gain meaningful traction—despite its potential to orchestrate complex, multi-step workflows.

So what’s holding it back?

Agentic AI isn’t just another AI tool—it’s a workflow revolution. Unlike generative AI, which excels at content creation, or predictive AI, which forecasts trends, agentic AI operates at the orchestration layer. It doesn’t just analyze data or generate text; it can execute decisions, coordinate actions across systems, and adapt to exceptions—all while maintaining compliance and audit trails.

For GBS, which manages finance, HR, supply chain, and IT processes for multiple business units, this capability is transformative. But it requires more than just plugging in an AI agent. Enterprises must first map their processes, audit their data pipelines, and define clear use cases before scaling.

What’s the playbook for real-world adoption?

Industry experts outline five critical steps

  • Map existing processes. Complex workflows—like those in logistics, where manual interventions slow down operations—must be documented before automation can replace them. A global shipping firm with seven GBS centers, for example, identified over 80 high-touch processes before reengineering them for AI.
  • Assess data readiness. Agentic AI needs structured data, APIs, and governance frameworks. Unstructured data or siloed systems create bottlenecks. Enterprises must ask: Where do data flow? What security and compliance risks arise when AI interacts with them?
  • Pinpoint pain points. Manual-heavy tasks with high error rates or compliance risks are prime targets. The shipping firm’s workflows, for instance, suffered from regional variations, SLA breaches, and legal exposure—problems agentic AI could address.
  • Test with an operating model. Pilots should align with real-world constraints. Options include centralized Centers of Excellence, citizen-developer approaches, or outsourced Build-Operate-Transfer models. Without structural clarity, even successful pilots risk staying isolated.
  • Scale incrementally. A multinational bank in Australia automated non-core processes first, then expanded to high-complexity workflows using an over-the-top platform. Within 14 months, it completed over 100 discovery projects—proving that scaling requires both technology and organizational alignment.

What does agentic AI look like when it works?

At scale, it’s not just about automating tasks—it’s about creating ecosystems where AI agents collaborate. Take procurement: document AI extracts data from purchase orders, but an agentic AI can also evaluate vendor risk, check compliance, verify budgets, and even negotiate terms—all while logging decisions for audits.

In financial advisory, predictive AI spots trends, but an agentic AI can assist advisors in real-time, suggesting strategic investments tailored to specific business units. The key difference? Agentic AI extends human judgment rather than replacing it, ensuring faster, more consistent decisions at enterprise scale.

GBS holds the key to enterprise-wide transformation. Unlike other departments, GBS sits at the intersection of processes and data across finance, HR, supply chain, and IT. This central role makes it the ideal launchpad for agentic AI ecosystems.

Standalone automation is a starting point. True impact comes when agents work together—sharing insights, learning from each other, and optimizing outcomes across the organization. For GBS leaders, the question isn’t whether agentic AI will reshape operations, but how quickly they can build the foundations to deploy it effectively.

One thing is clear: the hype cycle has passed. The work of making agentic AI a reality has only just begun.