Global business services (GBS), a centralized capability providing shared services across business functions, is at an inflection point. As agentic AI moves beyond rule-based automation to support judgment-based work, GBS has an opportunity to expand its role from delivering shared services to becoming the operational backbone of the enterprise AI transformation. Capturing that opportunity will require organizations to rethink workflows, talent, and operating models to unlock the full value of agentic AI.
In this episode of McKinsey Talks Operations, host Christian Johnson is joined by McKinsey Partners Heiko Heimes and Josh Peters. The following conversation has been edited for length and clarity.
Christian Johnson: Agentic AI is bringing us to an inflection point in how work gets done across the board. Can you paint a picture of what’s fundamentally changing in global business services?
Josh Peters: Global business services has been a hotbed for automation for many years. The difference we’re encountering now with agentic AI is its ability to act independently, understand context, adapt, and make real-time decisions. This is going beyond rule-based, individualized decisions and starting to apply judgment, pulling together multiple different contexts and a lot of information.
The potential for AI is really to become much more of an individual contributor and to reimagine how work gets done. Because of this, you might expect that GBSs would be shrinking, and we do see that in some areas, there is a lot of automation being applied. But we also see in other areas it’s really expanding: more people, more scope, more budget, more spend. So AI isn’t replacing GBS; it’s actually augmenting it.
Heiko Heimes: The truly successful GBS organizations fully embrace this continuous improvement, operational excellence mindset as part of their culture and behaviors. Agentic AI is simply the next step on that journey. If not GBS, there aren’t many other places in most organizations where you can build a foundation to really leverage that evolving AI capability.
Christian Johnson: Heiko, how is the addition of agentic AI affecting the scope of what GBS organizations can do?
Heiko Heimes: AI allows GBS and other parts of the organization to move more into advisory, strategic, and judgment-based topic areas. It’s moving beyond the more transactional, administrative processes, such as procure to pay in procurement and record to report in finance, and toward areas like strategic procurement, forecasting, and planning. These are activities where AI can play a significant role in assisting and also taking over portions of work. This is where GBS can play a much bigger role than in the past. The last dimension of expansion is that the frequency of activities can increase. In the past, we might have seen internal controls or auditing occur irregularly or only on a limited sample. You can now use AI for real-time processing, for example, in your internal controls, continuously checking transactions for compliance and regularity. Again, GBS could take on more work thanks to AI.
Josh Peters: What’s interesting, Heiko, is that we see these two things kind of countervailing each other. While a lot of work is being replaced by AI, more work is coming into the system. The question on everybody’s mind is, which of these forces is going to win? We believe GBSs are likely to be 20 or 30 percent smaller but dramatically more productive than they are today.
Christian Johnson: In thinking about how agentic AI can help increase the scope of what the GBS organization can do, Josh, you had an example from financial planning and analysis, or FP&A, that would illustrate the potential here. Could you share that?
Josh Peters: A great example of this is within the FP&A process area for finance. The classic issue is, how can we accelerate and automate reporting? GBS, for a very long time, has had a strong place in consolidating reporting work and applying more advanced analytics to it. Automation has helped to do that more and faster.
The interesting twist is how you move beyond answering a question to thinking about a next-best action. What we’ve started to see clients supplement is not just about year-over-year or month-over-month sales, why an area of the business is down, or why profit is down. Rather, it’s what should I do about it? When we’ve helped clients develop a copilot for FP&A, it’s not just saying, I think this is down because of seasonality or there was a weather event, but also, here are the two or three next questions you should ask, or the two or three next actions you should take.
Subscribe to the McKinsey Talks Operations podcast
Heiko Heimes: These planning-related activities are sometimes the most obvious area where suddenly GBS can play a more significant role, because AI helps GBS to really think ahead. Josh talked about FP&A on the financial side of planning and forecasting. You can translate this into HR, where we’ve been talking about strategic workforce planning, which organizations have long struggled to master. AI really makes a difference here, and there’s no reason why GBS can’t do the type of work that the rest of the organization hasn’t mastered in the past. Think about supply chain planning: You could go on and on, thinking ahead and planning next steps and actions. This is where AI gives you assistance, if not independent advice, and GBS is the perfect place for that.
At the same time, this means capabilities within GBS need to change. In the past, we might have seen organizations that are focused on more transactional, administrative-type activities. In the future, this needs to evolve right into more advisory, strategic activities. That’s something my clients are in the middle of thinking through, working out the steps required to make it happen.
Christian Johnson: That’s a great lead into the next topic, talent. That implies a very different, or certainly an evolving, talent model. What are the top considerations for organizations now in GBS?
Heiko Heimes: When we think about the future state of talent in GBS, we think about it as a diamond-shaped setup. This means you will have more in the middle layers of the GBS organization because we’re seeing less need for managing and running transactional activities. Automation—in the more classical sense, robotic process automation [RPA] plus AI—will do a significant chunk of the work. It will shift toward analytical, decision-making work that requires different skills and capabilities in addition to being able to manage AI, especially agentic AI. In addition to the general skill set that allows you to run planning processes within finance or strategic sourcing on the procurement side, you also need people who can manage the agent force, the set of AI agents that will take over in multiple areas within GBS. It’s a move away from the pyramid and toward a diamond-shaped talent model. That’s what we expect in the coming years.
Josh Peters: In particular, I think that trend means we have to move a lot faster to train and upskill entry-level talent, because we won’t have as long a cycle for that layer of talent to develop technical expertise and move into the supervisor layer. If a lot of that layer of talent goes away, we’re going to have to come up with new mechanisms to train people in different functional areas and prepare them to orchestrate and manage digital talent.
Heiko Heimes: It will also more fundamentally change the career paths we see in the organization, and that goes way beyond GBS. Today, a lot of GBS organizations see very high attrition levels, which means you constantly need to rehire, retrain, and find ways to make the organization as attractive as possible to keep attrition to a decent level. This will be an even bigger challenge going forward, because with less entry-level talent, you need people from the rest of the organization, outside of GBS, to come in, and perhaps people from GBS to go out. So, the collaboration between GBS and the non-GBS part of organizations needs to increase so that career paths make sense and are attractive to people.
Josh Peters: The lighthouse example I’ve seen is in the customer order and account management space, for both inbound and outbound customer contact. One company developed incredible technology for this; they knew the where and the why of the call, they could really understand what would make the customer happy, and they struggled to get the sales force to adopt it. I think there was a lot of fear: Is this really changing the way my job works? In our work, we learned a few things.
One is that the people side of the equation is as big, or even bigger, than the technological side. There was some work on training the models better, but I think a big part of it was, can you show them how this gets to better outcomes for both them and the client, versus can you just show them, well, we can divert 50 percent of the calls, and therefore we’ve driven a lot of savings. On the flip side, how does it enable them to do something more interesting or better with their time? Can they actually use augmented AI to get to a far better outcome, something that normally would have been escalated to somebody above their level, but that they’re now helping to resolve? Because they have the augmentation, the help, and the script that nudges them in the right direction based on what the customer says. Two, it’s just a more gratifying experience for the employee to work on something that used to go to their supervisor.
Christian Johnson: Since this isn’t an “or” but rather an “and” decision, how do organizations strike the right balance between what they’ve been doing with traditional GBS levers and how they apply AI?
Josh Peters: For a long time, we have seen a right-shore-first element or strategy for how GBS organizations think about migrating and improving their work. There is no one-size-fits-all answer, but we think there are two approaches. We’ll continue to see many organizations pursue right-shore first, which involves getting the work in the GBS for talent, for scale, for standardization, for capturing value quite quickly, and then driving further improvement through the application of AI once the work’s been standardized. We think that will be particularly useful for organizations that are seeking to accelerate value capture.
In parallel, we see an automate-first pathway, especially for organizations thinking about building a long-term digital capability in certain areas and more core strategic processes, where the question is how to unlock a lot of enterprise value.
Understanding and completely reimagining a process flow before migrating some of that work into GBS will be a high-value proposition. But the important thing is it’s not one-size-fits-all; it depends on the situation. Some parts of a process may benefit from moving to GBS first, while others may benefit from automating first.
Heiko Heimes: There are a number of factors when deciding which path to take. It starts with your own experience: What is your current setup within GBS, or with GBS in general? What’s your experience in leveraging AI and technology to improve processes? It’s also about your expectation in terms of impact: When do you need to achieve what? The impact can be cost-efficiency benefits, creating broader value, higher-quality outputs, more stable operations, and so on.
Finally, it depends on the type of processes. Some organizations naturally start where they already have GBS quite involved. So again, think about procure to pay, order to cash, record to report, if I stick to the finance domain. All these processes differ in how AI can be more or less easily adopted. So I think that combination of factors actually needs to be considered for the automate-first versus the right-shoring GBS-first approach.
Christian Johnson: What are examples or client contexts you’ve seen where the decision is being directed by some of the factors you’ve identified here?
Josh Peters: We’ve repeatedly heard: Shouldn’t we just always automate first? Is GBS even still relevant? Is there even a relevant right-shore-first path? In client situations where we see a greenfield approach—where they really don’t have a lot of shared services today, there’s a strong urgency for value unlock, a lot of low-hanging fruit, and processes that aren’t incredibly standardized and mature today—we are still absolutely seeing people choose the GBS-first or right-shore-first path to unlock value.
Where people start to deviate from that and think about automating first is when they identify use cases for unlocking enterprise value. The idea is, rather than eliminating some head count, reducing some process cost, or increasing throughput of a process by a little bit, are there other levers that allow me to reduce the amount of revenue leakage I have, improve the amount of price realization I have, and squeeze out a few more basis points of EBITDA? That’s where we see a lot of clients choose to automate first, because of the value unlock.
Heiko Heimes: I think the beauty of the AI discussion we’re having today: We’re moving away from a mere efficiency debate—such as how do I get cost reduction done quickly—to a more holistic set of ambitions. I always think about the three E’s: efficiency, effectiveness, and experience, as in customer experience.
Take the order-to-cash process, where we do a lot of our work when it comes to leveraging AI in a GBS context. This is about running the process more efficiently, so having fewer resources involved. But it’s also about creating a better outcome regarding, for example, revenue leakage, making sure people pay on time as originally aligned. It is also a customer-facing process, so if you do it right and run it seamlessly and smoothly, you can increase your customer experience levels, thereby creating much more bonding between you as an organization and the customer. It’s about really optimizing all three E’s and creating a massive impact for the company that goes beyond reducing a couple of heads or reducing a bit of cost.
Christian Johnson: We’ve talked a lot about the potential for AI to be transformative. What is being widely adopted right now? What’s the state of play for your clients, thinking especially of your more advanced clients?
Heiko Heimes: In my client portfolio, everyone is thinking about and using AI in the GBS context. Roughly half have started piloting and working on it, and leveraging AI to create the first use cases. The lighthouses I’ve observed in the past nine to 12 months—in the time when AI has started to make a difference—have happened in the more classical finance process domains and where the end-to-end processes are being allocated.
I have one client that used GBS for a long time in the backend of the source-to-pay process. The typical invoice comes in, needs to be processed, checked against the purchase order, and eventually paid, often in interaction with the suppliers.
The client adopted AI for the backend and connected it to the front part of the process, so when the purchase requisition comes out, there’s a conversation with preferred suppliers, and a purchase order is created. This organization has reimagined the process so that AI takes over a sizable portion of the supplier identification and selection, which is now being managed by GBS. GBS is more in charge of the front end, which determines how smoothly the back end of the process runs.
This gives more accountability to GBS and creates less friction, making the process more seamless. In this case, it also affects the external counterparts, the suppliers, who are more satisfied with how the overall process runs and how they get paid on time and as planned.
Josh Peters: An important thing to consider is not just getting started and testing things out. I have a client who asked us to kick the tires on their AI strategy, and a lot of it was: What is the set of ten or 20 use cases in each area that we think might have the most potential, that we may drive the most ROI from? I think it’s very important to set a North Star, and not just experiment but actually think about what value you’re going to create. What is the domain that has the most business value? Is it order to cash, as we were talking about earlier? They have to think about the different interventions across that process, some of which may be heavily AI-enabled, because the ability to reach out to customers can be done in a much more automated way, given the judgment that agentic AI can apply. But also, what other interventions are needed for a process that is not AI-enabled? As a first step, I would ask, do we have the right materiality threshold in place for how many of these things we pursue, and is there any low-value work we can eliminate? Otherwise, you’re still just automating the wrong work.
Christian Johnson: That’s actually a good call to action. You’ve talked about not automating the wrong work. What other messages or actions would you mention?
Heiko Heimes: Bringing the GBS and the AI aspects together. In my view, GBS should be an accelerator for any organization’s AI journey. When we think about service operations more broadly, GBS is typically the one place where you already have a good tech foundation. It’s where, in the ideal case, you already look at processes more end to end, not in functional silos. That gives you the foundation to fully reimagine how you run certain things.
That’s the basis you should build on: Bring it together, leverage GBS as an organization and as an entity within your broader company, and then see what AI can do. Don’t start too small or too large. Find that sweet spot in the middle where you can create a lighthouse without overburdening your organization with too much complexity.
Josh Peters: I remember the 2017–18 era, right after [RPA] got mature, when a lot of people, including all the consultancies and system integrators, said GBS is dead. I hear a lot of that sentiment now, people making proclamations that agentic AI means GBS’s time is short. I think the death of GBS is greatly exaggerated. It can still be the center where a lot of this process reimagination and end-to-end workflow orchestration can live, and GBS has the capabilities to do that. Ensure you’re pursuing both sides of the coin: the next-gen levers to apply AI and reduce work, as well as some of the classical levers to think about what more scope can be moved into GBS. It should always be a two-way street, and that’s the way we see successful GBS organizations move away from just transactions and cost and toward outcomes and value.
