The future of work is agentic

(upbeat music) - Happy holidays to you from us here at McKinsey. - Today, we've got one of our most popular interviews from 2025. We'll be back January 8th with new episodes. (upbeat music) - This is the McKinsey Podcast, where we help you make sense out of our world's toughest business challenges. - Welcome to the show, I'm Lucia Raheli, and I'm Roberta Fassara. (upbeat music) - I think we are gonna go into the world in which you will have to think about your workforce as both agentic and human workforce. - That's McKinsey, senior partner, Jorge Amar. He's talking about a future where humans and AI agents are sitting side by side in meetings. He spoke with talent experts, Brooke Weddle and Brian Hancock and Global Editorial Director Lucia Raheli. They dig into what AI agents are, how they can be used, and what that means for the structure of your teams. What you're hearing is a recent episode of their discussion on the McKinsey Tox Talent Podcast. (upbeat music) - Jorge, welcome to the show. - Thank you very much, excited to be here. - Jorge, there was a great little piece a month or so ago in the Wall Street Journal called, everyone's talking about AI agents barely, anyone knows what they are. Let's get to the bottom of that. What exactly do we mean when we talk about agentic AI? - I'll start where I think most people are still, which is generative AI, which is a combination of your favorite LLMs, large language models supplied, mostly for generation of content. I am sure everyone has tried generating that funny image, the cartoon style or something like that, but it is mostly a reactive type of AI that is focused on generating creative content mostly, triggered by a prompt or an instruction from an individual. That's where we are. Now, if we continue on the evolution of AI, now into agentic, we started to come into it very different reality, and the differences, the first one is, we're talking about AI that is not only generating content, it is doing something. It is really executing on a task, on a mandate, on a particular instruction. It can be proactive in the sense that it is not only now you are asking to generate an image, an AI agent is perceiving the reality, based on the training that agent has got, it then decides, applies judgment, and decides to execute something. And that execution, that final step that it takes, then reinforces the learning of the agent. So it is doing something, learning if what the agent did was good or bad, and then feeding that back in. So we are getting into the next step, we're not only now generating content for fun, AI is deciding what to do on its own, we start to get into this complete AI workforce. So if your AI agent could be now, the evolution and creation of a digital replica of the entire workforce that a company and organization has. - Okay Jorge, you're scaring us. Let's talk through some use cases for agentic AI that might help bring us to life a little bit. What does agentic AI actually look like now in the wild? - It's still the wild west out there, but I'll try, right now there are many companies that are starting to experiment with these, many more are going to come. Typically the environments in which they are deploying agents are very deterministic, that have a clear process to follow. So think of this as IT help desk, as IT or software development, customer service, tickets. So any of those environments in which there is a customer can be internal or external, they are asking for something, then there is a well-defined process that comes right after. The agent picks it up, decides what is the right process, the right content article to be retrieved, the right information to be gathered, and then triggers an action. So those are typically the ones that we are seeing come up right now. - In the world of HR, what we're seeing on agentic AI is in talent acquisition. So we're seeing agents clean records and go through and try to understand more precisely of the vast universal potential candidates. How do we clean the data and understand who might be the right candidate? And then having a separate agent go through and score those candidates and do the ranking and the sourcing process. And then a separate agent go through and actually reach out and try to gain contact and schedule interviews. And then I've seen a coordinating agent that sits on top of the overall process that's interacting with those underlying agents. Have you seen that kind of coordinating agent process and how do you even create an agent that coordinates across some of those discrete subprocesses? - You highlighted a number of potential agents when it comes to talent attraction, recruiting, reach out. I have a client of mine that is already doing the first screening of all the candidates for frontline, entirely with AI, with agents. I have even seen one step further, which is AI agents being deployed for training. So think of these as call center environment or store environment. You generate an agentic customer and that agentic customer, this is a type of call, just record an interaction. This is a type of customer and just train the agent. So it's not only simulating, as if it was a real phone call, you are also getting live as an agent, detailed scoring of how you are doing in that interaction. Are you using the right words? Have you remembered every single step of the process? So it gives you very detailed coaching instructions, probably before supervisor and a call center could listen to three, five calls per agent. Now you get a summary of every single call for the call center agent. With a detailed breakdown of all the things that this agent is doing, well, this human agent and all the things that this agent could do better. So you focus your coaching, your nesting, your onboarding in a much more targeted way because now you know exactly which are the skills to develop exactly the traits that you need to emphasize for these colleagues. It's not only that, it's the recruiting, the training, and you could even go once a further and do the exact same thing for performance management. Jorge, it sounds like you're pointing to examples where agents and AI have allowed companies to achieve greater levels of productivity. And I was struck, the work trends index annual report came out one of Microsoft's big flagship publications and workforce, and it found that a third of executives are considering using AI to produce headcount in the next 12 to 18 months. But nearly 50% said that they were considering maintaining headcount, but using AI as digital labor to what I interpret that as boost productivity, right? And use AI as complementary to human skills. What have you seen in terms of the use cases? As you said, it is still early days when it comes to what I call the decoupling of the creation of capacity that a tool like these provides, which is basically you are automating tasks that otherwise would have been performed by a human. There's creation of capacity and then what I call the monetization of capacity and the monetization of capacity as its own independent thing, which one of the potential paths can be, I'm gonna reduce headcount. I see more and more some of the executives that I'm talking to are interested in, I might reduce headcount, but I also might wanna do things differently because now with AI, let's take some of the companies that I work with, my competitive advantage, were my call center agents that were better than many of my competitors or my store employees now, if AI brings everyone to the same parity level, how do you differentiate? And what are the implications for your workforce where you can differentiate by just having the best algorithm and the best agentic framework out there, but at the same time, how do you complement that with humans to do things that otherwise would have been cost prohibited? I'll give you one example, I was discussing last week with one of my travel clients, and you could say pick the airline or cruise line of your choice, what if now you had your own personalized concierge that is looking at your travel, is giving you very detailed recommendations on how to navigate the airport that you're getting into would suggest the type of food that you can pick up on your way and even create the order for you and then even get to the final point in which it helps you board the plane and make sure you have space for it. So the possibilities are endless when it comes to figuring out or creating new different workflows, new processes, new ways to surprise and delight your customers that you couldn't have otherwise. And I imagine you can also do some of that same towards your employees, how do you surprise and delight across the employee journey? How do the agents actually get created and get created in a way where it's pretty specific to the processes in any one given area? That's a great question and something that I think we're all still trying to figure out what is the best way. There was a quote that said something along the lines of like IT will be the HR of the future. I would divide the creation of an agent in a few different steps that helps us understand who is doing what. The first one is there clearly needs to be a business rational that would only come from the business. Customer support marketing sales HR. They will define what is needed for an AI capability. Well, that's why what are the parameters of what this AI capability needs to perform? And therefore then they would work with their IT or AI function to either develop or procure their agentic capabilities. There are many cases in which the specificity and the complexity of the AI capabilities that are needed will require these companies to actually develop in-house their agent capabilities. Why? Because they cannot find it in the market. And it's going to be like a hybrid situation. Now, once that capability exists, comes what you would do with your human workforce. Now with your agentic workforce, you have to onboard and train that agent, which we call tuning of an agent. The tuning of the agent requires a number of things, a good articulation and understanding of the process you are trying to agentize. If you allow me the use of that word, you can agentize the process. You need someone that understands that. So an SME that really understands the ins and outs. The second one is you need someone that understands the available data. So you need a content specialist that is saying, these are the content articles, the corpus of knowledge that you need to train your agent on. And make sure that they are up to date. In one of my cases, we trained the agent and the agent started to spit out a bunch of COVID-related policies that were no longer relevant for the situation in which my client was operating right now. So you need to make sure that the data is accurate, it's relevant up to date. And last but not least, if you need a good robust prompt engineering skillset, someone that can teach and train and tune the agent by saying, hey, when the customer or your employee says this, this is what they mean. This is what they are trying to accomplish. And therefore do X, Y, Z. Jorge, you mentioned IT will become the HR of AI agents. And of course, it was Jensen Huang, the CEO of NVIDIA, who said this recently. When you think about a digital workforce, whose job is it to ensure that that digital workforce is reaching its full potential? There's all sorts of directions you could go in. Is it more in the realm of IT in terms of making sure that the digital workforce and specifically agents are reaching their full potential? Or is this a space where HR actually might have a few things to say, knowing that for a long time, getting managers to reach their full potential has been more in their purview? How do you think about it? I would start by saying, hey, there are some of the pioneering companies in this space that are expressing their org charts, not only in number of FTEs, but also number of agents that are being deployed in every part of the organization. So I think we are going to go into the world in which you will have to think about your workforce as both agentic and human workforces. This is my own personal take. I don't think IT will be able to do this alone. I think IT will be critical in enabling the foundational elements to train an agent, the data stack, the right platform for training and tuning the agents. Now, the two missing pieces, one is the business. Nobody will be able to train an agent if you don't know, intimately, the policies, the processes, what really differentiates you from a business perspective. And then I think HR will play a key role. The first one is really push the business on what can be done from a hybrid workforce perspective. The second one, we started seeing these in one of my clients where the technology was up and running. The number of live interactions was not coming down. And there is a big change management component that will need to come into play. And I think HR will be absolutely critical in doing that. How do you tell your 20-year tenure employee in the call center or in your back office that now there is this agent that is going to do the job much better than you? This person will probably say, how can this AI thingy that got trained yesterday going to replace my 20 years of experience? For example, there is a big step towards driving the incentives for usage, the role modeling of the communication, the right change story for this employee. It's saying like, hey, your job was this, but look at all the great possibilities that this unlocks for you. So this tells me HR still will play a critical role in the adoption of this agentic workforce. Change management will be even more critical. So maybe HR will not be screening each resume. For a call center position, but it will be critical in making sure and driving the change management efforts in the call center for adopting an agentic AI workforce. Jorge, it's so interesting to me to hear how the kind of anthropomorphic terms you're using to describe these agents the way that they exist in the org chart, for example, or describing them as a digital workforce. To be clear, are these agents being construed as tools or as a class of digital worker, like neo-colleagues of some kind? I do think of it as a workforce. And this is of a workforce that will conduct end-to-end processes, will in many ways replace the tasks that are being performed today by the human workforce. It will augment the tasks that a human workforce is performing to help them make it better, faster, more efficiently. There are some companies out there that are even promoting this notion of like, hey, we will get to a zero FBE department in many companies. And therefore, an entire function fully performed by an agent. Then you have on the side, humans in the loop controlling or monitoring what these agents are doing. So putting the philosophical debate aside, I think we should think of them as a parallel workforce for all intents and purposes. And we so often hear that adoption is a primary challenge in realizing the value of AI generally. How do you see humans in the workplace taking to this notion of collaborating with AI agents? So it's still a big challenge. I'll give you one example. In some of the front-line environments that I spend a lot of time on, some of the newer agents or the newer reps, then to embrace AI faster. Why? Because if you're just coming into a front-line environment, a back office, where you need to learn all these things. And now AI is guiding you through the process. That's great. It makes my job easier. Some of the more tenured employees resist AI quite a bit. And it's really challenging for them. The other big element that we see when it comes to adoption is there are many employees that tell us, I cannot trust an AI black box out there that is doing this. So I will use the AI result. But at the same time, I'm going to have my own calculations. And therefore, you're now duplicating work. But there are many of these elements that I think are going to be critical in cracking the code for adoption because my fear is that we will end up with huge investments and very little value realized. Who do you think is going to lead the way in adoption? First, there's got to be a clear mandate from the top that is using it and making sure that they are role modeling and integrating AI into the way they speak. They do and they perform. In one of my clients, for instance, they are seeing the results of both the human part and the agent part of the operation in the exact same dashboard. And the business manager, the PPE, the SVP, they are evaluating the joint performance of both their workforces. Second, you evaluate the performance of AI in a joint fashion. This is intimately ingrained in the way that the operations are running their day to day. Third, this space is changing week by week. Day by day. And therefore, you would need to design an operating model of a set of processes that allow you to adapt the more flexible you design, this operating model to be the better. Because otherwise, you're going to be making investments on a technology or a set of algorithms that three months from now are going to be different. So if you put all that in the mix, some of the smaller companies startup environments have a little bit of an advantage, for sure. The reality, some of these LLMs or agent platforms are not going to be trained on small companies. So it is critical to get to the larger companies to say, hey, I'm going to make the performance of this even better. And therefore, how to do that in that environment? That's to me the cracks of this issue. How do we do that in an effective way? What would be the skills that are going to become more salient in human leaders to get the most out of agents? The first thing is that HR will need to be at least business proficient in what an agentic workforce can do, because how can you drive a change management program if you don't know what your agentic workforce is able to do and what it cannot do? Or what will be possible in three years? Number two is, I think HR will play a pretty important role in the reskilling of a number of the existing human employees. Today, you can probably fully agentize the workload of a level one in support engineer. But you might want to repurpose that person to become a prompt engineer or to go to content generation for AI training. So an HR function that can do that at scale, that is to me another critical component and skill so that HR will need to develop if you think about the next three, five years. What is the evolution of the role? And the last one that I would point out is around just being really good at empathy, understanding of the change story and helping employees onboard into their own AI journey and make it in a way that is not threatening to them. Look at all the other possibilities that you might have in the future within the organization, articulating that very clearly and helping employees come along in that journey is going to be another critical component. You know, that Work Trends Index report that I talked about earlier, that same report talks about the need to evolve from an org chart to a work chart. Yes, and you probably have seen that the CEO of Shopify released a memo a few weeks ago as well saying something along those lines, which is before you ask for your head count, show me that AI cannot do the work. That was almost positioned as a more radical stance. But in my conversations, that is very much part of the conversation already, right? And so I very much think that's a now thing versus a future thing. Now, if you allow me, I think there are a couple of elements that we also need to put on the table to say, why now or why not now. And I would describe them into three broad categories. Number one is to get an agent up and running, I would say, you do need a good technology stack and data stack. And there are many things that are being done to create new data, generate what we call synthetic data for training purposes. Number two is that are a number of concerns about any of the security and risks that are out there from an agent perspective, from drift hallucination by any of the challenges that you have with some of these LLMs where what if you have an agent interacting with your end customer and these customer starts to gain, or even worse, you have an agent talking to your customer support agent and they generate its own little dynamics and negotiation. And now suddenly you ended up with a 90% discount on your product because you trained your agent into churn reduction and churn avoidance. And now they found a way. So how do you control that? And maybe you need to train a whole new set of agents that are monitoring the different negotiations and the different discounts and anything that touches your CRM. And the third is what is the cost? What are the different usability considerations also from a UX and UI perspective? It's great you might have a very conversational chat pod but if it looks like the 1990s interface of how you were interacting on some of your most famous messaging platforms, customers are not going to use it. So I think it is a very now conversation but it also requires to tackle some of these issues around risk, data, usability, because otherwise it's going to go into a purgatory. That's not where we want to go, clear. On this question of now, obviously it's vital to be talking about this now, planning for it now, but acknowledging that predictions are freighted with uncertainty. What kind of a time frame do you think we're talking about for agents really to take effect at scale in companies? Depends on who you ask. Some of the hyperscalers and technology companies would tell you that they are already deploying it and they are. Many of the other organizations that I talk to are saying, "Hey, this I need to understand, I need to evaluate." And we're probably looking at 18, 24 months out before it reaches full scale. I will be probably very wrong six months from now and I will not get another invitation to come talk to you guys on this. But I do believe that there are a few elements in which it's going to take a little bit of time making sure everyone is comfortable with deploying them at scale. So Jorge, I've got two college age kids and what I'm wondering is as they're looking to go into the workforce where we've got these agents and it's not an org chart, it's a org chart and playing into the future, what advice do you have for them as they're thinking through their careers and how to engage and work in the future that is agentic? I was having this conversation if you would go with a friend's son that was asking me something very similar of like, hey, maybe I should just drop out of college and go become a prompt engineer. I was like, okay, well, maybe who knows? In the world of predictions, maybe I'm going to get that one wrong as well. Look, I think there are certain jobs that are going to be fully transformed by AI and these net new roles that I call like prompt engineer content specialists will become more relevant in organizations. There will be a lot of demand. I would expect this demand to be higher than what the market can offer when it comes to just college. And therefore, I think we will have to go through re-skilling at scale within the existing workforce. On the other side, how do you differentiate? And if you differentiate only by just having the best prompt engineer find, I think that is a skill set that is certain point you will catch up on because you could even have an agent that does prompt engineering. And if you think that the most important element that a company has is trust and relationship with their customers, do you need a human workforce that is more empathetic? Because again, you might be okay talking to a chatbot to re-skate your unemployment. But if you were just in an accident, do you want to talk to a human? Or do you want to talk to a bot? And therefore, how do you emphasize some of those skills in the incoming human workforce that make a company establish relationships with humans? If not, this will be a bot company or an agent company just interacting with humans. This could be the source of differentiation for your company. This could be the competitive advantage of like I offer a superior service. I offer a more human touch, surprise, and delight experience. And therefore, my friend's son, I was so opening his mind in terms of like, hey, maybe prompt engineering is fine, but maybe my arts background will be valuable in tomorrow's workforce because I will be able to understand human feelings in a way that no agent will be able to do. Jorge, if I just reflect more on what you're saying, I think it's a good time to also think about broader cultural implications of having a digital workforce. And some of that even relates back to the values of a company. As we think about incorporating and onboarding agents as part of an organization, how do you think about doing that in a way that is consistent with your company values where you might prioritize collaboration, psychological safety, or having the difficult conversation? I think it's a really interesting question to be asking to get full value out of the digital workforce. 100% and that's why I think we are seeing more and more companies starting to experiment with employee-facing agents, more than just full end customer-facing agents. Because how do you make sure that every interaction is in line with your corporate values, with your identity, with your brand standards, with your way that you want to address a customer? That's why I think we're going to go first through a testing and a scaling of an employee-facing agentic workforce. And then over time, in certain discrete moments, you might want to do it with your end customer. You might want to do certain tasks that are mundane, that are customer authentication, or verification, or call summarization at the end of it. But again, you don't want to outsource to an agent, to a digital agent, the core of their relationship with your customer, or not just yet. - Yeah, it's fascinating. I read the article you recently published, Jorge, about Agente AI in the context of customer care. And I found it fascinating that one of the findings there was that almost three quarters of Gen Z respondents to your survey actually believed that live calls were quickest and simplest. I wasn't surprised that, for example, boomers almost all of them want to talk to a human. But even younger cohorts seem to prefer that human interaction. So there must be a tremendous change management process that will have to be underway in order for this to take hold. - It is so funny, you mentioned that. For us, it was surprising to see the Gen Zs, they would be bothered if they got a phone call from their parents, right? And they would prefer to interact with them just by text message. I am sure anyone who has kids can relate to that. But when they get to their customer support needs, they prefer to talk. So when we were digging a little bit deeper into why, they prefer to talk to their provider, insurance company, telco carrier bank. They all mentioned the same. It's that my situation is so unique, so important to me. I just want to talk to a human that will give me that personalized and unique solution that I won't be able to get through about. And the reality is maybe 80%, 90% of those interactions, the human was just giving them the exact same process. But still, they felt they had a much more personalized solution by just talking to a human. So maybe that changes over time and your customer support bots will just get also two characters as a fully formed dialogue. So to your point, the change management is not only with employees, it's also customer education and you're building the trust on some of these solutions. - And Jorge, just more of a reflective question, when you think about the next three to five years, what are you most optimistic and excited about when it comes to the potential of a genetic? - I'm really looking forward to doing things in a way that we couldn't have done otherwise. I am super optimistic on doing personalization at scale with customers. I am super optimistic about putting humans to do tasks that are not repetitive, that are not gonna create attrition levels of 50, 60, 100% per year, and that we can develop as a society. We're a little bit more philosophical that we are able to create career paths for employees that are really about connecting with humans, transforming the way that we work on a daily basis, focusing on the change management elements that we were just describing. This hybrid workforce future should be a very uplifting environment for everyone, mostly for us, as part of the workforce, that it creates a new set of skills that were probably deprioritized in previous ways of finding efficiencies in companies, which is like you try to do this thing as repetitive and as fast as possible, and really open the door to new ways of interacting with customers and employees. Fascinating, this is the most interesting discussion. I've had in such a long time, I mean, it's just fascinating. - I think it's great, I have a fun question. Should we tell AI, thank you? - I do, because I think that when Skype net takes over, I want them to know that I was very kind to them. (laughing) It's funny, I was reading the other day in that open AI is spending tons of cycles. I don't know if you've seen that news of please and thank you. - I say thank you to Alexa. I mean, it's just a good behavior all around. - Of course. The article with Open AI was saying like we're spending millions of dollars pumping tons of CO2 in the atmosphere because of the energy used by the data centers, because we're saying thank you. - Totally, but imagine your kids, you're gonna teach them to not say thank you to Alexa, but say thank you to a human. - Come on. - Exactly, I mean. - Right, right. - Jorge, that was fantastic. Thank you so much for joining us. - Well, of course, thank you for the invitation. - Thank you, I learned a lot. - Thanks so much for listening to the McKinsey podcast. I'm Lucia Railey. - And I'm Roberta Fissaro. - Find us on McKinsey.com. We'll have a transcript of this episode up shortly. - And download the McKinsey Insights app where you can find this podcast in other helpful content updated daily. - If you enjoyed the show, we'd love for you to leave a rating and a review. - We'll see you in two weeks. (upbeat music)