EY helps clients create long-term value for all stakeholders. Enabled by data and technology, our services and solutions provide trust through assurance and help clients transform, grow and operate.
At EY, our purpose is building a better working world. The insights and services we provide help to create long-term value for clients, people and society, and to build trust in the capital markets.
Agentic AI in GCCs: From delivery hubs to intelligence engines
Explore the EY.ai Unplugged podcast on how AI-powered GCCs are evolving into intelligence engines driving smarter decisions and enterprise transformation.
In this episode of EY.ai Unplugged podcast series, we explore the evolution of Global Capability Centers (GCCs) from delivery hubs to into intelligence engines that enable faster decisions and smarter execution with Sunil Venkatesh, Partner, Technology Consulting, EY India. Sunil shares insights on autonomous AI agents that are reshaping enterprise workflows, governance and talent strategies and why leaders should strike a balance between innovation and control.
Key takeaways:
GCCs are uniquely positioned to lead AI transformation, combining enterprise data, process expertise, scale and governance maturity across functions.
Agentic AI is transforming GCCs into execution engines that automate workflows, reduce manual effort and accelerate enterprise decision‑making at scale.
The future GCC will focus on intelligence ownership, measured by decision speed, resilience and AI governance.
New AI‑driven roles will emerge, including agent designers, AI workflow architects, supervisors and governance leads who will ensure responsible execution.
AI first is not about removing humans, but placing them where it matters most, which includes moving them from execution to judgment, oversight and accountability.
Sunil Venkatesh
Partner, Technology Consulting, EY India
For your convenience, a full text transcript of this podcast is available on the link below:
In today's episode, we are joined by Sunil Venkatesh, Partner, Technology Consulting, EY India and together we will unpack how Global Capability Centers (GCCs) are evolving from cost- focused delivery hubs into AI-native engines of innovation, intelligence and strategic decision making.
We will also explore what it truly means to adopt the AI-first operating model, how organizations can responsibly scale autonomous agents, and how leadership teams can navigate the balance between innovation, control and trust in increasing autonomous enterprise environment.
Hi, Sunil. Welcome to the podcast. It is a pleasure to have you here. Could you help us know how is Agentic AI transforming GCCs from cost savings centers into AI-native intelligence and decision-making hubs for enterprises?
Sunil
Thanks for having me. The biggest shift is that GCCs are moving from being places where work gets done cheaper to becoming places where the enterprise learns how work should be done differently.
Historically, GCCs started out as cost arbitrage hubs. Then they became global business services platforms. And now they have become innovation or tech hubs. Agentic AI takes them into a fourth phase, where AI is not just a tool that is sitting on top of existing processes. It becomes the execution layer of the enterprise.
And this is very different. Traditional automation says let me make this more efficient or let me make this task faster. Agentic AI says, let me understand the objective, plan the steps and call the right systems, resolve exceptions and escalate only when needed. So, instead of a person moving work from one system to another, an AI agent can orchestrate workflows across finance, supply chain, HR, compliance and technology.
And in that sense, GCCs are uniquely positioned because they already have scale, process knowledge, enterprise data access, governance maturity and engineering depth, which is a rare combination in enterprises. A business unit may understand its own processes, but a GCC is often able to see an enterprise horizontally.
In simple terms, old GCCs helped enterprises save money; new GCCs will help enterprises think, decide and act faster, which is a slightly better job description than ‘please do this cheaper.’
Pallavi
How does an AI-first operating model look like in practice, and how are autonomous AI agents reshaping enterprise workflows and governance?
Sunil
An AI-first operating model does not mean sprinkling AI on top of every process and hoping some magic will happen. It is more like putting a spoiler on a bullock cart. It may look modern, but it will not transform your performance.
In practice, AI-first means you redesign the process around what AI agents can do natively. The workflow moves from linear handoffs to event-driven orchestration. Instead of the work waiting for the next person, the system continuously ingests data, detects exceptions, makes recommendations or decisions within guardrails, and triggers the next action.
Let us take finance as an example. In the traditional model, teams wait for period and data. Then they reconcile accounts, case exceptions, prepare reports and escalate issues. In an AI-first model, a finance close agent continuously ingests ledger data, detects anomalies in real time, performs reconciliations, prepares drafts and surfaces only the exceptions that truly require human judgment.
Same applies to supply chain. Instead of weekly planning cycles, agents can read demand signals, check inventory, understand lead times, coordinate with logistics systems and recalibrate plans continuously. The rhythm changes from batch and review to sense and respond. Governance also changes. You cannot govern autonomous agents the same way you govern a spreadsheet macro. You need explainability, audit trails, access controls, human override and model monitoring.
The operating model has two sides – more autonomy in execution, but more discipline in governance. What would best describe this is AI-first is not about removing humans from the loop, it is about moving humans to the right part of the loop – from repetitive execution to judgment, supervision and accountability.
Pallavi
As organizations scale Agentic AI, how should leaders balance innovation with risks around accountability, oversight, and human-in-the-loop decision making?
Sunil
Leaders need to avoid two extremes. One extreme is: let a thousand agents bloom, which sounds very innovative until one of them wires money to the wrong vendor or sends a regulatory filing to the wrong regulator.
The other extreme is to wrap every AI use case in so much approval that innovation dies quietly. The right balance is controlled autonomy, and that starts with classifying decisions. Not every decision needs the same level of human oversight. Low risk, reversible, high-volume actions can be highly automated, but decisions with material, financial, regulatory, customer, employee or reputational consequence need ‘human in the loop’ or ‘human on the loop’ governance.
I would like to give you a good example. Think if you are riding a horse and you are taking it on a road. There will be times when the horse takes the decision, and there will be other times when the road is narrow or there is some obstacle in front, where the rider is the right person to make the decision, and the agent is, in this sense, analogous to the horse.
We have to think about where the agent should be allowed to make decisions, and where the human should take over and make the decision. Leaders need to think about innovating very aggressively but also think about architecting responsibly. Give agents room to operate, but put them inside a well-lit room with cameras, guardrails and off switches.
Pallavi
How will the rise of Agentic AI redefine talent strategies inside GCCs from workforce skills and organizational structures to new AI orchestration roles?
Sunil
I am sure you have also read all the newspaper headlines about how AI will make jobs disappear, and that is the least imaginative version of the conversation.
The more important question is what new work becomes possible when people are no longer trapped in repetitive process execution. Inside the GCC, the workforce model will shift from doing manual work to designing, supervising, improving and governing how AI agents do the work. That requires a very different skill mix. We will see roles like AI Workflow Architect, who redesigns processes for agent orchestration; agent designers and prompt engineers who configure how agents reason, interact with tools, and respond to exceptions; agent performance supervisors who monitor live deployment and manage exception queues and responsible AI governance leads, who will ensure agents operate with ethical, regulatory and enterprise risk boundaries. These are just a few of the job titles that will get created.
The broader workforce will need AI literacy. Finance professionals will need to understand how reconciliation agents will work. Supply chain planners will need to understand how demand sensing agents will work, and risk teams will need to understand model behavior, not just policy documents.
Organizationally, GCCs will become more cross-functional. The old model separated by process, technology, analytics and risk will change and the Agentic AI workforce will force you to group that work into what I would call an intersection where the agent sits on top of the process logic, data, systems, controls and user experience.
Smart GCCs will be the ones that will redeploy productivity gains into higher value work because frankly, removing all the people who understand the process and then asking AI to improve the process is like firing a chef and asking the oven to redesign the menu. That is not going to happen.
Pallavi
What will differentiate the next generation of high-impact GCCs from traditional delivery centers in an increasingly autonomous enterprise landscape?
Sunil
This goes back to the first point that we made, which is that the next generation high impact GCCs will be differentiated by one big thing – they will not just run processes, they will own the intelligence architecture behind those processes.
Traditional delivery centers are measured on cost, throughput, service levels and efficiency. Those metrics still matter, but they will not be enough. High impact GCCs will be measured on decision velocity, enterprise resilience and quality of AI governance, speed of scaling, reusable agentic capabilities, and their ability to create value beyond efficiency.
The differentiators will include four things. First of all, Agentic AI native talent and culture – people who can work with AI systems, challenge them, supervise them and improve them. Second is platform architecture, which is modular, API-first and allows interoperable systems of agents to work together. Third, of course, is governance by design, where we need to have auditability, explainability, risk controls, and human oversight built into the workflow, not as an afterthought. And fourth is operating model redesign – moving from hierarchy and handoffs to real-time orchestration and continuous optimization.
So, the future GCC will not be a back office or even an innovation hub. It will be closer to an autonomous intelligence center, a place where human expertise and agents work together to help the enterprise sense faster, decide smarter, and execute with more confidence.
Pallavi
Thank you, Sunil. Now that brings us to the end of this episode of EY.ai Unplugged Season 2: AI Agents of Impact. A special thank you to you once again for joining us and sharing all your insightful perspectives.
To all our listeners, if you have joined this episode, do follow the series for more conversations with experts who are shaping the future of AI driven impact.
Thank you for listening. Until next time, this is Pallavi signing off.
