Where AI is Driving Revenue Growth Today
22:03 Share

Today on the AI Daily Brief, the latest McKinsey survey on AI, including where companies are actually starting to see business value accrue. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes.

Hello, friends. Quick note, today's main episode went very long because there's lots to talk about with this new survey. And so there will just be a main episode today, no headlines. We will be back with the headlines in our normal format tomorrow. For now, let's dig into McKinsey's latest state of AI report, with this one coming out about a week and a half ago. Now, in terms of how to think about reports like this, I would never ever look at them as the gospel truth on exactly where companies are.

In an incredibly high-moving, high-velocity field like Gen AI, the normal variance that you'd see across organizational surveys is going to be even more heightened. And so instead of viewing this as anything definitive, view it as a strong piece of evidence to help you understand where AI is when it comes to big companies. Now, I say big companies, but actually, this is a pretty wide cross-section of people who were interviewed.

Unlike the KPMG survey that we talked about a couple of months ago, for example, only 42% of the respondents of this survey worked for organizations that had $500 million or more annually in revenue. So there are some smaller companies as well. What's more, the leadership level of these respondents varies more widely than some of the other surveys we've covered as well. There are a bunch of interesting insights in here that I think in some cases reflect our experience with what we're seeing at Superintelligent.

spending all of our time talking to a similar range of different types of companies about their AI strategy and AI deployments. But then there are some other aspects of this that show, I think, just how far behind companies are, even in this area where they're particularly attuned to trying to stay up to date.

Let's kick it off at a high level. To the surprise of no one, McKinsey found that Gen AI use has accelerated meaningfully over the last couple of years. The percentage of organizations that use Gen AI in at least one business function has gone from around 35% back in 2023 to 71% when this survey was conducted. Interestingly, where Gen AI is being used shows a really wide distribution.

On the bottom end, you have manufacturing and supply chain and inventory management, which only see 5% and 7% respectively of respondents using Gen AI for those functions, all the way up to the top, which is marketing and sales at 42%. Product and service development is at 28%. Information technology at 23%. Service operations at 22%. Knowledge management at 21%. Software engineering, 18%. Human resources at 13%. Risk, legal, and compliance, 11%. Strategy and corporate finance at 11% as well.

Couple things that stand out to me about this. One, overall, these are still really low numbers. And this will be a big theme, I think, throughout this conversation is that these numbers are just lower than you might expect. A couple that really stand out to me, only 11% of these respondents using Gen AI for strategy and corporate finance seems crazy to me. But then again, this may be an underappreciation of how good AI is as a brainstorming partner when it comes to strategy. But the other big glaring one to me, see if you can guess it, only 18% using Gen AI for software engineering. Now,

Now, this is a theme we'll come back to, so I won't dwell on it here. I also think it reflects something around when the survey was conducted. But still, even though we have seen over and over again a surprising lack of adoption of AI by software engineering departments inside these big companies, just 18% is so ludicrously low it almost boggles the mind.

Now, there is some variance by industry, but not as much as you might expect. In other words, across industries ranging from technology to professional services to media and telecom to energy and materials, marketing and sales was the biggest use case across pretty much all of them. Maybe the most notable thing when you look by industry is which industries in general are leading versus lagging. For example, just 8% of CPG software engineering groups are using Gen AI right now.

At the same time, they have a comparatively high percentage of their marketing and sales groups using AI, but everything else is really low. On the other end of the spectrum, technology and media and telecom are using AI much more fluidly throughout their operations, indexing higher across a lot of different categories compared to other industry peers.

Still, even in the technology sector, only 36% of software engineering groups were found to be using Gen AI. To put that clearly again, only one third of the software engineering departments of technology companies were using Gen AI. Insert some trope about how early we are here, but it's true.

Maybe more important, though, I think, than the growth in organizations using AI in at least one function is the growth we're seeing in organizations that are using AI across multiple functions. All of these numbers are up significantly to the right.

Over the previous period of McKinsey's research, the percentage who were using AI in two or more functions went from 50 to 63. The percentage that were using it in three or more functions went from 27 to 45. The percentage that were using it in four or more functions almost doubled from 15 to 28%. And the percentage that were using it in five or more functions did double from 8 to 16%. Low overall, but it does feel like once organizations start to lock in on one function, they may grow into more functions more quickly.

Next up, let's talk about where the risk focus is. And to some extent, what's interesting to me is not the exact areas of risk focus here, because in many ways, this is exactly what you'd expect. The top risk areas that organizations are concerned with and are working to mitigate are inaccuracy, cybersecurity, and intellectual property management. Regulatory compliance and personal and individual privacy are a little bit behind that.

But if you look at the trend lines of where there is growth in focus of which risks organizations are working to mitigate, this tells the story of a Gen AI practice that is so, so much about deployment.

The risk areas that companies are most focused on working to mitigate are all the ones that increase as you actually do more stuff. Meanwhile, and this is something we'll come back to a little bit later, the percentage of organizations who identify workforce and labor displacement as a risk they're working to mitigate has actually gone down to under 10%. Point being that this tells the story of an enterprise space that is very, very focused on actually deploying AI and getting it into market.

Now, where are things from an organizational process? It's one thing to say that X percentage of companies have AI deployed in at least one area, and another to actually try to understand the current state of organizational processes that allow them to use AI more fluidly.

One overarching note is that the bigger companies were more mature than the smaller organizations. Now, on the one hand, this makes sense. They have more resources to do things like establish dedicated teams. But at the same time, I think you could argue that smaller organizations have more of an opportunity to be nimble than their bigger peers and competitors. And so to some extent, this reflects to me a lost opportunity for those smaller organizations.

In any case, what are some of the organizational processes that McKinsey asked about? They found that 52% of big organizations had established a dedicated team to drive Gen AI adoption versus just 23% of smaller organizations. Now, I wonder again how much this reflects the fact that in some smaller organizations, instead of a team, we're going to see an individual. And so maybe there's a bit of semantics here. But still, I would say both only 52% of big organizations having established this and only 23% of small organizations seems very low.

and certainly lower than we're seeing at Superintelligent, although again, we're going to be biased towards organizations that have this sort of thing because those are the organizations that are most likely to come to us for help figuring out where they stand and finding the right partners.

When it comes to establishing a clearly defined roadmap to drive adoption of Gen AI, only 25% of big organizations have done that versus just 12% of small organizations. When it comes to creating a feedback mechanism for the performance of Gen AI solutions, 24% of big organizations have that versus 19% of small organizations. And when it comes to tracking well-defined KPIs for Gen AI solutions, just 18% of big organizations and just 16% of small organizations.

Now, broadly, I think this reflects the fact that Gen AI adoption has been important enough that the lack of clarity around KPIs hasn't been slowing anyone down so far. But still, these numbers are really, really low. And it definitely feels like something is going to give here. In fact, as I've argued before, I tend to think that part of the interest in agents is that they sort of make some of these questions of KPIs and ROI more broadly, a little bit easier for organizations to wrap their heads around. In that if an agent works, which of course is a big if at this stage,

Working implicitly means doing the thing more efficiently or for lower cost than the comparative human equivalent. Overall, these numbers fit with a companion study that McKinsey released back in January, where they found that just 1% of companies that were investing in AI believed that they were actually at a mature stage with it. And of course, when it comes to maturity, what companies are ultimately going to be interested in is where Gen AI is actually being used to make money or reduce costs.

On the reduced cost side, some of the biggest decreases came in HR, supply chain and inventory management, service operations, software engineering, and IT.

In fact, in HR, 15% of respondents said that they had used GenAI to decrease costs by over 20% over the previous 12 months. In both supply chain management and service operations, while the overall totals were high, 46% and 45% of organizations saying that they had used GenAI to reduce those costs over the previous 12 months, that was largely concentrated in a decrease of costs of less than 10%. Today's episode is brought to you by Vanta. Trust isn't just earned, it's demanded.

Whether you're a startup founder navigating your first audit or a seasoned security professional scaling your GRC program, proving your commitment to security has never been more critical or more complex. That's where Vanta comes in. Businesses use Vanta to establish trust by automating compliance needs across over 35 frameworks like SOC 2 and ISO 27001. Centralized security workflows complete questionnaires up to 5x faster and proactively manage vendor risk.

Vanta can help you start or scale up your security program by connecting you with auditors and experts to conduct your audit and set up your security program quickly. Plus, with automation and AI throughout the platform, Vanta gives you time back so you can focus on building your company. Join over 9,000 global companies like Atlassian, Quora, and Factory who use Vanta to manage risk and prove security in real time.

For a limited time, this audience gets $1,000 off Vanta at vanta.com slash NLW. That's V-A-N-T-A dot com slash NLW for $1,000 off.

Today's episode is brought to you by Super Intelligent and more specifically, Super's Agent Readiness Audits. If you've been listening for a while, you have probably heard me talk about this. But basically, the idea of the Agent Readiness Audit is that this is a system that we've created to help you benchmark and map opportunities in your organizations where agents could

specifically help you solve your problems, create new opportunities in a way that, again, is completely customized to you. When you do one of these audits, what you're going to do is a voice-based agent interview where we work with some number of your leadership and employees to map what's going on inside the organization and to figure out where you are in your agent journey.

That's going to produce an agent readiness score that comes with a deep set of explanations, strength, weaknesses, key findings, and of course, a set of very specific recommendations that then we have the ability to help you go find the right partners to actually fulfill. So if you are looking for a way to jumpstart your agent strategy, send us an email at agent at besuper.ai, and let's get you plugged into the agentic era. Hey listeners, want to supercharge your business with AI?

In our fast-paced world, having a solid AI plan can make all the difference. Enabling organizations to create new value, grow, and stay ahead of the competition is what it's all about. KPMG is here to help you create an AI strategy that really works. Don't wait, now's the time to get ahead. Check out real stories from KPMG of how AI is driving success with its clients at kpmg.us slash AI. Again, that's www.kpmg.us slash AI.

Now, back to the show. When it came to revenue increases, the areas that saw the highest gains were strategy and corporate finance, supply chain and inventory management, marketing and sales, service operations, software engineering, and product and service development. Now, one of the things you might remember is that when we were looking at which areas were using Gen AI overall, strategy and corporate finance, supply chain and inventory management, these were on the very low end of the spectrum.

So there seems to be a disconnect between how much value organizations that are actually deploying Gen AI for those areas are finding versus how many organizations are deploying Gen AI for those areas so far. Supply chain and inventory management was the area that saw the biggest revenue increase of more than 10%, with 19% of respondents saying that they had seen a revenue increase of more than 10% from using Gen AI in that area.

Now, keep in mind, just from a methodology standpoint, only people whose organizations were using Gen AI in these areas were asked how much Gen AI was increasing revenue in that area. So, for example, the 76% of respondents that say that Gen AI increased revenue in supply chain and inventory management, that means 67% of the something like 7% of organizations that were actually using Gen AI in that area.

Now, with all of these different functions having different profiles when it comes to AI, it brings up an interesting question of how AI is being managed. Is it being managed in a centralized way or a decentralized way? In other words, from some central leadership or leadership committee or more at the group department or line of business unit level?

Well, it turns out that different areas of AI have different centralization profiles when it comes to management. And this one I think is pretty intuitive. Areas like risk and compliance and data governance tend to be more centralized. 57% of respondents, for example, had fully centralized governance around risk and compliance. On the other end of the spectrum...

Things like specific adoption, tech talent, and even building out roadmaps in AI strategy had a little bit more of a balance between decentralized and centralized strategies. Adoption of AI solution, for example, found 23% was fully distributed. In other words, all resources are living within the business functions, which matched the percentage that was fully centralized, i.e. organized by a hub or center for excellence, but the vast majority, 54%, was hybridized in some way.

Now, one of the things that's interesting that we're seeing is the rise of agents is definitely pushing more rather than less decentralization. For the first couple of years of AI, when a lot of the adoption decisions were about things like governance structures or were about picking which major platforms, Microsoft Copilot, Google Gemini, OpenAI, et cetera, a company was going to go with, they wanted centralization of that decision because they didn't want one business unit using Gemini while the other business unit was using Copilot.

agents are introducing a re-balkanization of that buying process, which is sort of more reflective of how things were in the SaaS area, where you're not necessarily going to see the same companies providing the best solutions across all the different business units.

So I would expect to see the decentralization of adoption do nothing but increase as agents become more and more of the focus. One other note that I thought was really interesting as it related to governance, even though they didn't have a chart for this, they write basically that CEO involvement is one of the most correlated areas to companies saying they're seeing financial gain from AI. McKinsey writes, a CEO's oversight of AI governance is one element most correlated with higher self-reported bottom line impact from an organization's Gen AI use.

That's particularly true at larger companies, where CEO oversight is the element with the most impact on EBIT attributable to Gen AI. One of the things we talk a lot about here is that when it comes to adoption, one of the key dimensions is leadership and communication from the top. These results from McKinsey seem to verify that.

Now, a couple of things that seem off to me about this. So far, a lot of what we've talked about has pretty well reflected our experience at Super Intelligent and just my experience more broadly talking to companies kind of day in and day out for this show.

One of the areas that really stood out across this as just crazy to me was how low software engineering and coding were indexing as use cases for Gen AI inside these companies. For example, while 63% of responding organizations said that they were generating text with AI and 36% said that they were generating images, only 27% said that they were generating code.

Given how much software engineering is being changed in front of our eyes by these text-to-coding tools right now, this seems just remarkably low. But at the same time, there is at least one bit of an explanation here. Keep in mind, this survey was just published on March 12th, and yet the survey was actually conducted from July 16th to July 31st.

Now, no shade to McKinsey because this is valuable research, and I'm glad that they do it, but we are no longer in the paradigm where you can realistically survey people in July of the previous year and expect it to give you really meaningful signal around what's happening right now in March of the next year.

That's effectively a third of the time we've had since ChatGPT was launched. And in this specific case, we see just how problematic this is, as there are three major categories of innovations that are not represented in this data. One are the reasoning models, which only started to come out in the fall. The second are the text-coding tools and coding agents that also started to become really popular in the second half of last year.

And third, you might notice that there's been no mention of agents because agents have only been a going concern for the last three to six months. As a snapshot of a picture in time, this is still really valuable, but it is out of date before we're even talking about it.

which either means that McKinsey needs to catch up and move faster with this stuff, or it means that there's an opportunity for someone else to swoop in and be a more definitive source of information by running their own research like this. Anyways, when it comes to this coding and software engineering thing, given how many of the exciting tools that people are using now really hit product market fit in the period after this survey was conducted, it maybe makes sense why these numbers are so low.

At the same time, I think even with that, we would find, if we were surveying right now, a shocking resistance to adoption on these areas from the software engineering groups inside these big companies. This is something we find over and over again, and I think, frankly, is going to be very challenging and problematic for those companies if they can't get their software engineers on board.

Now, speaking of reskilling expectations, let's move over to the area where McKinsey asked what share of employees had been reskilled in the past year due to AI and what percentage expect to be reskilled over the next three years due to AI. The percentage that were reskilled over the past year was low, but maybe to be expected. Only 9% said that more than 50% of employees had been reskilled because of AI, whereas 44% said that up to 5%, in other words, a very small percentage had been reskilled due to AI.

Another 40% or so saw between 6% and 40% of their organization having been reskilled in the previous year. Where I think organizations are critically off is in their expectations of share of employees that are expected to be reskilled over the next three years. Certainly, we do see an increase across all categories. 19%, for example, say that they thought that over 50% of their people would be reskilled, and about 60% thought that between 6% and 40% of their employees would be reskilled over the next three years.

Now, if you've listened to me at all, you'll know that my prediction is that 100% of organizations will see more than 50% of their employees reskilled over the next three years due to AI. And I'm only choosing over 50% as the top indicator because that's where McKinsey ended things. I think the fact that only a fifth of these companies think that more than 50% of their workforce is going to be reskilled based on AI means we are still living in a moment of denial relative to the actual magnitude of this change.

And the big question, of course, is how this is going to impact headcount. I believe that the central question that will shape how disruptive to society Gen AI is from a workforce perspective will be what percentage of companies choose to view AI just as an efficiency technology where they can output the same with fewer inputs versus what percentage of them see AI as an opportunity technology where they can keep the same workforce or even increase it a little bit to see and capture massive new opportunities that were never possible before.

In this McKinsey study, 38% of organizations who are using AI said that they predict Gen AI will have little effect on the size of their organization's workforce in the next three years, which is either a good thing in that they're fully appreciating just how big an impact AI is going to have and are excited about the opportunity to create and build more rather than cut headcount, or that 38% reflects a denialism that doesn't really understand just how disruptive AI is going to be.

Of course, we're not going to be able to know that until we get a little bit farther down the field and we see how these companies actually handle the changes that are coming for them.

One positive thing that McKinsey found, they write, when it comes to the headcount impact of AI, C-level executives are more likely than middle managers to predict increasing headcount. That means that the people who are in charge of strategy are perhaps a little bit more biased than their middle management peers to view AI as the type of opportunity technology that I'm talking about rather than just an efficiency technology. I don't think there's enough signal there to put too much stake in it, but it's certainly more positive than if they had found the reverse.

Anyway, overall, despite the fact that this survey was conducted in the ancient years of mid-2024, I still think there's a lot of value here. Hopefully, if you are a listener who's thinking about these questions for your enterprise, there's some useful comparative data. I will, of course, include a link to this study in the show notes. And if you are interested in some more direct benchmarking and opportunity mapping of AI and agents inside your organization...

shoot me a note. That is exactly what we do at Super Intelligent, agent and AI benchmarking, opportunity mapping, and then matching you with the right providers who can help. For now though, that is going to do it for today's AI Daily Brief. Appreciate you listening or watching as always. And until next time, peace.