AI will automate most of what GBS does today. But AI will not automate what GBS should become: the governance infrastructure on which every enterprise AI deployment depends and the engine room for top-line growth in the enterprise.
For three decades, GBS has defined itself by what it processes: transactions consolidated, cycle times reduced, service levels maintained. That model delivered genuine value. It also created a ceiling. AI removes the ceiling by automating the work—and in doing so, forces a more consequential question: what is GBS actually for?
The answer, for organisations willing to design for it rather than drift into it, is something far more strategically significant than a cost centre. Every AI deployment in the enterprise requires clean data, governed workflows and clear accountability. Those are GBS's capabilities—and they are precisely the infrastructure that AI depends on to scale reliably. The function built on operational efficiency is now positioned to own the intelligence layer of the modern enterprise.
Why AI intensifies business services
AI does not eliminate the need for business services. It intensifies them. Intelligent systems depend on clean data, governed workflows and clear accountability. The more intelligence an enterprise deploys, the more it requires disciplined data stewardship, integrated automation and consistent standards. These are precisely the domains GBS has spent thirty years building.
What changes is the scale and pace. Transaction volumes compound. Data flows multiply. Exceptions become signal-rich rather than simply inconvenient. An organisation running AI models across finance, supply chain, and HR needs coherent governance for all of them—who deployed them, how they are performing, where their outputs are used, and what happens when they are wrong.
That governance capability does not emerge automatically. It must be built and owned. In the AI enterprise, GBS is not made redundant. It is made essential—if it chooses to be.
Efficiency is now table stakes
Historically, GBS proved its value through cost reduction and service-level performance. Headcount efficiency, cycle-time reduction and standardisation were the core metrics.
AI changes the baseline.
Process mining tools surface bottlenecks automatically. Intelligent automation handles reconciliations and validations. Generative AI assists knowledge work in finance, HR and procurement. The Hackett Group's 2025 research found that 42 per cent of GBS organisations had already piloted generative AI, with 63 per cent of early adopters reporting measurable gains in productivity, cost savings and service quality. These are not experimental outcomes. They are early operational baselines.
In this context, efficiency becomes the expected outcome of deploying modern technology. It is no longer the strategic edge.
Performance leaders in GBS already run at 29 per cent lower G&A costs than their peer group—a $110 million annual advantage for a $10 billion organisation. AI does not create that advantage. It assumes it. The competitive question moves upstream: what does GBS do once the efficiency floor is established?
If GBS defines its mandate purely as cost compression, AI commoditises it. The differentiator shifts from processing transactions to governing intelligence.
"Efficiency is the expected outcome of deploying modern technology. It is no longer the strategic edge."
The strategic choice
Enterprises now face a structural fork. In one mode, call it reactive convergence, AI capability grows organically. Business units adopt tools independently. Vendors embed intelligence within ERP, CRM and workflow platforms. Automation proliferates within functions. GBS adapts to what emerges.
The outcome is not dysfunction. Capabilities expand, costs fall, and individual pilots deliver genuine value. But architecture is emergent rather than designed. Data standards fragment. Governance follows deployment rather than preceding it. The enterprise becomes a passenger in a journey whose direction is set by vendor platforms and internal experimentation.
In the other mode, call it proactive design, leadership defines the architecture. Data standards are owned at enterprise level. Automation platforms are consolidated and governed. Model risk frameworks are established before scale. AI investments are explicitly aligned to commercial priorities. GBS becomes the operational steward of the shared capability layer through which intelligence flows.
From supporting the business to powering the engine
Traditional GBS supported the business. AI-enabled GBS can power it. Powering means enabling growth as well as efficiency. An enterprise capability engine operates shared demand forecasting models that improve revenue capture, enable dynamic pricing insights across markets, orchestrate working capital optimisation across procurement and finance, integrate supply chain and commercial data to accelerate speed to market and provide enterprise-wide analytics services that inform strategic decisions in real time.
These are not back-office outcomes. They are competitive levers.
Unilever's UniOps function illustrates the trajectory. With more than 500 AI projects deployed and 23,000 employees trained in AI use, Unilever now processes 240 terabytes of data weekly across its supply chain network, executing more than 13 billion computations daily. Predictive inventory models deployed across 100,000 freezer cabinets delivered sales uplifts of 8 to 30 per cent across markets. These outcomes did not emerge from deploying AI in individual functions. They came from governing intelligence at enterprise scale—precisely the mandate that GBS is positioned to own.
In this model, GBS becomes a driver of top-line performance, not merely a guardian of cost.
What an AI-Led GBS operates
An enterprise capability engine is not a slogan. It has structural components:
- •Enterprise data stewardship. The foundation. AI cannot scale reliably where data quality is inconsistent or ownership is unclear. GBS is positioned to own and enforce the standards that determine whether enterprise AI produces reliable outputs or accumulates error at scale. This is an executive question, not a technical one.
- •AI and automation platform operations. The shared infrastructure through which intelligent capability is built and maintained. Shell's GBS function, operating more than 100 AI applications across upstream, downstream and integrated gas, anchors this on a data lake processing trillions of rows—governed through a framework that prioritises elimination and standardisation before automation.
- •Model governance and risk oversight. How AI models are procured, deployed, monitored and retired. This is not a technical function. It is a control function—one that carries direct financial and reputational consequences as AI becomes embedded in operational workflows.
- •Shared analytics and insight services. The analytical capability domain teams need, without each team rebuilding it independently. The cost of intelligence falls when it is shared. The quality of intelligence rises when it is governed.
- •Interface capability. The human layer that translates between AI systems and business decision- making—the professionals who define standards, own consequences and align intelligence with strategy. Automation executes. Humans govern.
This is AI-enabled, but human-governed.
The agentic horizon
Generative AI was the first wave. Agentic AI—systems capable of executing multi-step workflows autonomously across enterprise systems—is the second, and it arrives faster than most GBS organisations are prepared for.
As of 2024, only 10 per cent of organisations had deployed AI agents in production. But 82 per cent plan to do so by 2027. In GBS terms, this is not a distant horizon. It is the next operating cycle.
An agent resolving a supplier invoice query may simultaneously access ERP, contract management, procurement records and payment systems—taking action, not just generating recommendations. An agent supporting payroll reconciliation may cross-reference HR, finance and compliance systems in minutes, flagging anomalies and initiating corrections autonomously. Early implementations are already compressing process cycle times by up to 70 per cent.
For GBS, the implication is structural. Agentic AI does not merely automate individual tasks. It executes workflows. The governance infrastructure required to deploy it safely—data standards, model oversight, accountability frameworks, audit trails—is precisely what a well-designed GBS is built to provide.
Organisations that have deferred that design conversation will find themselves deploying agents without the architecture to manage them. The risk accumulates at machine speed.
"Agentic AI does not merely automate individual tasks. It executes workflows—and the governance infrastructure to manage them has to exist before deployment, not after."
What this asks of leaders
For COOs, this reframes GBS from an efficiency programme to a growth enabler. Funding shifts from incremental cost optimisation to platform investment. Governance shifts from SLA monitoring to enterprise capability stewardship. Talent shifts toward data governance leaders, automation engineers and cross-functional translators who can connect intelligence to commercial outcomes.
For CFOs, the implications are equally structural. Investment decisions move from cost-centre budgets to enterprise infrastructure. Return on investment must be evaluated not only in headcount reduction— though that matters: Hackett Group research shows top-performing GBS organisations achieve $10.30 in value for every dollar invested in AI—but in decision quality, revenue uplift and risk mitigation. Governance of AI becomes a financial control issue, not a technology programme.
The talent question deserves specific attention. The Hackett Group identified AI talent shortages as a barrier for 67 per cent of GBS organisations attempting to scale, alongside process complexity (73 per cent) and data quality issues (71 per cent). These are not technology problems. They are design problems. And they require leadership decisions, not vendor solutions.
Four shifts are consequential:
- •Treat data governance as a strategic priority, not a technical function. Data quality determines AI quality at scale. GBS that owns data governance owns the quality of enterprise intelligence. The executives whose decisions shape enterprise data are business unit leaders, CFOs and COOs—not the technology function.
- •Establish the governance model before tools multiply. Most organisations are making procurement decisions faster than governance decisions. The architecture of accountability needs to precede the architecture of capability.
- •Invest in the interface layer. The binding constraint in most enterprises is not access to AI tools—it is the capacity to connect AI outputs to business decisions. That requires a different kind of professional: cross-functional, analytically fluent, commercially oriented.
- •Measure outcomes, not deployments. Tools deployed, pilots completed and efficiency targets hit are incomplete proxies for value. The measures that matter are decision quality, revenue contribution, risk reduction and speed of iteration.
Business services capabilities will expand in every AI-enabled enterprise. The technologies ensure it.
The inflection point is not whether GBS will be transformed by AI. It will. The inflection point is whether leadership shapes that transformation deliberately—or inherits whatever emerges from fragmented adoption and vendor-driven defaults.
In the AI enterprise, scale of transactions is no longer the defining challenge. Scale of intelligence is.
GBS can remain a support function optimising efficiency. Or it can become the enterprise capability engine that governs how intelligence is built, deployed and compounded at scale.
The difference lies not in technology. It is a design choice—and it is available now.