Loading summary
Transcript38 lines
- [00:00]
A
Foreign. Welcome to the AI to ROI podcast, the Big Story episode. I'm Ray Reich. I'm founder and CEO of BenchMarket.
- [00:19]
B
And I'm Peter Buchanan. I'm the managing partner of New Plan.
- [00:22]
A
And this is the weekly big story episode that Peter and I do together, where we dive in to what we think is the big story of the week, from our AI to our newsletter. And we do five editions every week. And this week we picked something that I was really enamored about. So, Peter, what are we going to be discussing today?
- [00:42]
B
We're going to be discussing a report on AI in 2026 from Deloitte that's super comprehensive. It concentrates in the enterprise, it looks at the C suite, all the way down to individual workers. And it could be really helpful to companies that are really diving into making their AI projects go into production this year.
- [01:07]
A
Okay. And the nice thing is this is an annual report, so you also get to see some trends. It is a 41 page report, once again, the Deloitte 2026 State of AI report. And we covered that on our February 19th edition of the newsletter, which you can see at substack, that's AI to roi.substack.com and Peter, the report, they called it out this year the Untapped Edge. So, you know, first of all I thought about edge computing and then I'm like, no, I think maybe this is a polite way of saying that the survey found we have a massive inventory of untapped AI potential. Is that right?
- [01:48]
B
It is right. And there are sort of seven areas that Deloitte looks at as inflection points. The first is the one everyone talks about, AI moving from pilot and experimentation phase to the enterprise as workers get expanded access to AI and AI use cases. The second is transformation of productivity, reimagining businesses around AI, which is a big quest. The third is companies are focusing on AI fluency of their employees so that they have the skills to actually take advantage of these use cases and help the company. The fourth is of course, agentic AI. So we've had three years of generative AI getting better and better, but agents are here now. The fifth is sovereign AI because a lot of companies do business all around the world and they have to comply with where countries that where they do business. The sixth is physical AI, which is sort of snuck in. It's with all the devices and control systems you have. And lastly, it's how leaders are thinking about AI and there's a little bit of a shift this year. They feel like they understand the strategy and where they want to go. But of course they look at it and say, oh gosh, the devils are in the details. So let's get, let's get started, Ray, with just the first part here, the pilot to enterprise scaling part. What's the report say?
- [03:15]
A
Well, before we actually get into that pilot to full commercial scaling, I wanted to talk a little bit about just access at the individual level. Right. So the report says that workforce access AI has expanded by 50% year over year, growing from 40% of workers have sanctioned access. And that's very important. People have been using chatbots outside of the firewalls for a while, but it was 40% were sanctioned last year, 60% in 2025. So that's nice. 11% of leading companies actually are providing workers with near universal access to sanctioned AI tools. More than 80%. However, even though 60% of workers have access to, to sanction AI tools, only 60%. So if you do that kind of Math, it means 36% of employees are actually using AI in their daily workflows. So, and that hasn't changed. And there's some good details in the report about what percent of employees aggressively adopt it and what people actually kind of do the pocket embargo. So a key question for me was, okay, so we're getting better at sanctioning and approving tools. We have, you know, 60% of people have access to those tools. But I still remember that MIT report on are people actually implementing AI and proof of concepts or pilots and then are they converting that into full scale commercial enterprise wide deployment? What did the report say, Peter, about this?
- [04:56]
B
Well, I mean, the report said it's definitely the pilot to production is definitely the most important step. And companies are stalling and they struggle to scale and find measurable success. They've had that struggle all of 2025 going into 2026. But this year, enterprises seem to be more positive. So 25% of the respondents said that their organizations have moved 40% or more of their experiments into production. That's a big jump from the previous year. And then the other thing that's very interesting is that the respondents, over 3,300 respondents to the Deloitte surveys, 54% of them said that they expect in the next three to six months that they're going to get a huge leap in shifting pilots into production. So these respondents think their companies are ready, ready to go.
- [05:56]
A
Yeah, you know, they say that, but you know, also in the report, I don't have the data right on the top of My head. They've said they think that their strategy is well prepared. I think was it like 40% said their strategy was really well crafted for their AI journey.
- [06:16]
B
Right.
- [06:16]
A
But they just weren't operationally ready. So to me, it's like if you don't have a strategic plan that's linked to detailed plans of action, that is plans of execution, you don't have a complete strategy. Peter, that's just my belief.
- [06:32]
B
So you're an experienced business executive. You've experienced it before.
- [06:36]
A
Well, you and I worked together a while ago back when GE was, you know, one of the better management leadership places to work. And I still remember our CEO at the time said, you know, having a strategy is nice, but if you don't have a plan of action, don't talk to me about it. It's a little bit like Michael Tyson says, hey, everyone has a plan until they get hit in the face. Right?
- [07:00]
B
Right. That's right.
- [07:01]
A
But hey, let's talk a little bit about AI productivity and you know, is it really happening? So real world businesses impact is rising fast per their survey with 25% by 25% seems like a real common number here. But 25% of leaders are reporting that AI is having a transformative effect on their company and that's up from 12% year over year. So that's nice that people are self reporting that. And it also is reflected in investments because 84% of the organizations actually surveyed are increasing their AI investments. And 78% of the leaders report that their confidence in the technology is higher than it was last year. But here's some of the edge issues and data points. 66% say that they have increased efficiency and productivity today. Let's be honest, that's primarily individual efficiency and productivity. They. That wasn't called out in the report, but I've seen that 53% are saying, hey, it's helping us enhance decision makers, enhancing decisions. That's going to increase to 61% in the future. So a small lift. Only 40% say that it's about reducing costs, where 65% say that in the future and 38% say that today it's helping to enhance customer relationships or improve products and services, while 60% of companies say that's important in the future. But Peter, what I did notice in these facts I just pulled from the report is it didn't say much about revenue. Is it having any impact on revenue today?
- [08:43]
B
Yes. Well, you have the efficiency and productivity gains. You're, you're not getting that. It's lagging the other indicators, the back office is ahead of the front office. So 74% of organizations hope that they can grow revenue through AI initiatives this year. Only 20% of it of organizations are doing it today. So what this says is that AI is on the verge of breaking out and delivering go to market benefits. And success isn't about just efficiency or even growing revenue. It's about strategic differentiation. So bringing all of those benefits together so that you can begin to have a truly transformational effect in your company. So Ray, there seems to be some stratification here. Leaders, middle laggards, how are companies progressing?
- [09:41]
A
Yeah, right before I kind of go into that kind of next topic, I was just thinking about this revenue impact and to me, you know, there's a key question here. Is it about being more efficient and effective in your go to market, you know, customer acquisition, retention and expansion or is it more about new product introductions? And I think about another newsletter edition we did where we talked about three great case studies at Notion, Canva and ServiceNow. And Canva was really interesting where they've embedded AI into their existing product and it makes the people using it, the designers much more efficient and doing a lot more designs.
- [10:23]
B
So.
- [10:23]
A
So I think it's going to be important over time to say is it more efficiency and current with current products or launching new products. So just a point I had, Peter.
- [10:33]
B
Right. I think the other angle in revenue is, and not all companies have put this in yet is the products companies are using to support the sales cycle. A lot more of them are doing a great job of researching and aligning the sales process with, with the ICP of the company that's trying to sell the products. And so that you're going to see that infusion into the sales cycle this year in a major way because there's a lot of AI native and companies that have transformed into being AI driven that are including those capabilities now and salesforces are going to take advantage of it.
- [11:13]
A
It's all agreed. Now there was another thing here in the report and it really, you know, it jumped out at me. So for the listening audience, to me this Figure 3 and their report is really important and it really dealt with their approach to business transformation with AI. Boy, do you remember the days, I think it was like 10 years ago, Peter, we were talking about digital transformation. So we had to hire that chief digital officer, right. Not sure how much real business productivity we got of it because if you look at labor productivity growth over the last 10 years, we've kind of been a little lethargic, around 2%. But I digress. Let's go into this report. So Deloitte says that 34% of companies are redefining how they work to fully leverage AI. But let's really talk about that. So 34% are actually saying there's deep transformation of their products, their processes and business models, but 37% saying there's little or no change to existing process. So to me, until you actually go in, redesign your entire jobs, the work to be done, the end to end processes to optimize the leverage and integration of AI, you're not going to get optimal benefits. And only 30% actually are saying they are redesigning key process around AI. So, you know, in fact, I think you and I have talked about this, Peter. OpenAI just introduced their Frontier Partnership Program to have the McKinsey and BCGS and Accentures of the world help their enterprise customers who have tried to leverage OpenAI and Chat GPT to automate business process. And they realized that a lot weren't going into production, a lot weren't getting the roi and, and it needed change management and business process reengineering to do it. So, Peter, I know you're doing a lot of research on how companies are actually deploying AI. Do you have a story to share that's relevant to what I just shared regarding how companies are approaching these AI initiatives?
- [13:26]
B
I do. So Deloitte had an example of a head of automation and global engineering at a mining company and he talked about how they took a bold transformational approach by embedding AI into all of their core products. So they transformed traditional mining equipment into connected platforms with sensors and predictive analytics. That meant that, oh well, meantime, between failures coming up, or this equipment isn't quite working right right now, or it's time to replace these components before you have an adverse event. So here's the money quote. He said AI is much better than a technology. We wanted to give it to everyone for everyday usage and make it pervasive everywhere. So everybody says that all the time. But he ends the quote with but we also wanted to disrupt the market. So you'd think that'd be obvious. And also it's hard to do for most companies, but it's a big step. So companies are actually first concentrating on what they consider to be AI fluency, getting their employees to be very comfortable. Not necessarily these big leaps in business transformation. Maybe companies aren't putting enough into AI fluency though, right, Ray?
- [14:52]
A
Yeah, I don't know if it's AI fluency Peter, in fact, I'm going to go back to the fact that what I mentioned just a couple of minutes ago, that 37% of respondents say that they're having little or no change to existing processes when they're trying to use AI to transform, which makes no sense. But here's the oh my God. This highlights a real issue to me that 84% of companies have not redesigned jobs around AI. Now once again, I understand we're in the early, very early innings of deploying agentic AI to fundamentally transform processes, which means you're transforming the jobs that humans used to do as part of that process. But 84% haven't done that. Now, giving people access to AI to get experience, that's great. But to actually have that experience translate into fundamental business process reengineering, I know that's an old word, but that's fundamentally what needs to happen. You're investing in employee fluency without getting the business benefit. So here's another thing that really jumped out at me. I think about half of companies are making significant adjustment to their talent strategies. But their talent strategy, it's more about how their focusing on educating employees versus rewriting job descriptions and how they hire. I'm going totally off script. And a conversation with a Fortune 100 CEO that believes they can reduce their employee count by 50% over the next five years. Thanks AI. And when they have senior leadership positions transition, they go out and hire top MBA student to serve in that role. And their job in the first 90 days is to come back and share a plan on how they're going to re engineer that entire function and the business processes that result in material cost reductions. And that includes human cost reduction. So to me that's where we're going. But here's something that jumped out in the report. About a third, 36%. So just a little over a third. And expect at least 10% of jobs to be fully automated within a year. And I go immediately to does that mean that those 10% of jobs will be repurposed and reallocated to other roles or they disappear? Now I know we're going to be covering the Citrini report on one of the scenarios they said where we had unemployment of 10.2% at the end of 2028. But my question to you Peter, is are 10% of jobs going to be fully automated due to the adoption of agentic AI? Is that the key?
- [17:55]
B
I think it is the key. So companies want to deploy agents because agents set goals. They can reason through multi step tasks. They can use tools like APIs to access data, they can coordinate with people or other agents. And so that changes AI from a source of information and insights to a system that can perform in a different capacity. It can go end to end across departments inside a company. And so companies are anticipating major rollouts and advancements of AI, of agentic AI and AI agents. So what does Deloitte say?
- [18:37]
A
Well, in this report, and that's why I like the fact that they do this, I'll call it longitudinal. Now, they need to have multiple users reports. We're just talking year over year. But it says last year's report said 26% of organizations were exploring autonomous agent development. So agentic AI to a larger, very large extent. And this year, 23% of companies are already using agentic AI at least moderately. But the real key is project out to the future. And they asked two years out, how will you be leveraging agentic AI? And three in four companies, 74% to be exact, say they will be using it at least moderately. And 23% say that they're actually going to be using it extensively. So one of my questions here is, okay, that's pretty good. What will they actually be doing? What will these agents actually be doing? Peter?
- [19:34]
B
Yeah, so agents are expected to have the highest impact in customer support. And that's in fact the, that's in fact the number one use case so far. But there are a ton of use cases. So supply chain management, dipping even outside your company to reach suppliers and suppliers of suppliers, R and D to speed up development of new products, knowledge management to make workers smarter and to bring together disparate pieces to get insights in a way that's more efficient than generative AI. Cybersecurity, absolutely critical. So in your cybersecurity stack, 12 or 24 months from now, almost all of your every product you have is going to be an AI driven product. It's the only way to defend against attacks. So Deloitte gives some examples. Like in financial services, they have a client that's building agentic workflows, that captures meeting actions from video conferences, it drafts communications, it reminds participants of next trips, and it tracks follow through, which seems kind of creepy, but also makes things much more efficient. So an air carrier is using AI agents to allow customers to complete some very common transactions like rebooking flights or rerouting bags, so that human agents can deal with things that are more complicated. And so so far, agents aren't eliminating jobs, they are upgrading jobs because humans have to manage the performance of these agents and account for agentic risks. And in fact agents do create significant risks and risks are an impediment actually to the rollout of agents. So what does Deloitte say about agentic AI and risks?
- [21:38]
A
You know Peter, here we are, we're 25 minutes into our episode, right? We got five minutes. And I really do want to cover a couple of the key risks that I see from this report. And one of those is 21% of companies report that they have a mature model for governance of autonomous agents. Now once again, since only, you know, less than a quarter actually aggressively using agent AI, that makes sense, but it's still a low and worrisome number, especially in regulated industries, you know, financial services, health care, etc.
- [22:13]
B
So right.
- [22:13]
A
To me, developing those governance frameworks for AI has to be a top, top focus over the next few months and years. And I'll say part of that is also having AI explainability that audit trail. That's one of the big benefits of context graph where they can actually have audit trail and history of why a decision was taken, what policies were leveraged, what precedence was used, what regulatory and compliance guidelines were followed. So I think that's a huge issue and didn't know if you have anything to add to this kind of lack of a mature governance model yet.
- [22:54]
B
So. So yes, the companies that are having the most success with agents as are putting in governance models and there's they're building governance models but they're also starting with their lowest risk agentic use cases first, building the governance models around those use cases, scaling them, making sure they work and then they put in place cross functional governance structures. Because many agentic use cases go from product development to product management to marketing. They go across a company and so they have cross functional structures that have it legal compliance, business unit leaders, employees that interact with these agents all the time and they set up policies, they monitor performance, they manage escalations and they get better evolutionarily over time. It gives them more confidence to do more complex agentic solutions.
- [23:57]
A
You know Peter, I another thing we covered in our newsletter, the AI to RI newsletter available on substack to the listening audience was a little bit about the emergence of Chinese AI models and open source models. But when I think about governance and becoming more mature, I think about sovereign AI, especially in highly regulated industry. So can you tell me about how about that's kind of sovereign AI and local government. Maybe either current regulations or potential regulations are affecting how companies go about deploying agent Ki.
- [24:31]
B
Sure. So sovereign AI is when a country and the companies that operate in it, they design and train and deploy AI AI under their own laws on infrastructure, they control locally and data is stored locally. And the goals are to reduce dependence on foreign vendors and to keep critical capabilities in country and to build in country AI skills. So it's a big boardroom issue. Now 83% of multinationals, board members say that 83% of sovereign AI is at least moderately important and 43% say is it's extremely important. And so the companies across borders have to have, have to have plans for this. So what are they actually doing?
- [25:27]
A
Right, well let's provide some context around this. 77% of companies at least factor in the air solutions country of origin and their vendor selection and 58% now build their AI stacks primarily with local vendors. And I think it didn't say this report, but the assumption is where I'm seeing multimodal utilization. Some of the lower risk, less confidential information might be using some open source or out of country models. But the primary high risk has regulation are using in country models. But I know that there's some data to report. It's not the same everywhere. And we're coming up to the end of that podcast, but I do want you to just cover what areas of the world are probably taking sovereign AI a little bit more seriously.
- [26:21]
B
Yeah. So in the US it's not really a big issue. 11% of American, American companies rely on foreign source solutions for the majority of their stack, partly because the leading products have been developed in America. But in Europe and the Middle East, 32% have mixed stacks. So ultimately sovereign AI isn't about technology ownership, it's about strategic independence for the country. You know, it's having control viewed by their own data, their own models, their own talent, their own ecosystem. And these countries want to be able to innovate locally and compete in AI.
- [27:00]
A
Yeah. I'm going to wrap it up with one last data point I saw in the report and then a quick summary, then throw it back to you Peter for a quick summary. But the thing that jumped out at me was, and I've said this a couple times, so to be specific, 42% of companies believe their strategy is highly prepared for AI adoption, yet their operational readiness falls way below that. And 30% say the same about risk and governance. Both are increasing their preparedness since last year's report. But for me, in summary, here's what I'm thinking. Number one, we have to reduce the gap between employee access, company strategy and operational readiness for activation One of the things we got to do is redesign the work and jobs around an AI first mentality that we got to build that governance that needs to be part of the the proof of concept before you can scale the governance along with the proof points are critical. We do need to be very cognizant of those sovereign AI requirements so not let the technologists or the financial people looking for better models, cheaper models impact our sovereign kind of strategy and policies and that we need to build an intrinsic AI infrastructure in data model that lives and evolves for tomorrow. And it's more about pursuing strategic reinvention including the ability to drive more revenue growth. 74% of companies say they're going to be able to do that and not just focusing on incremental efficiency. So that's kind of a little bit of my summarization. Peter, any final things you'd like to wrap up today with?
- [28:46]
B
Overall, I'd say that, you know, we're, we're definitely at an inflection point. I thought the trends coming out of the Deloitte Report, despite this sort of give and take, we're positive if companies are getting a handle on what they need to do in the Future. I think 2026 will be another year of fits and starts. But when they write this report in 2027, they're going to see a lot of progress.
- [29:11]
A
I'm looking forward to it and to those companies out there. I'm kind of glad I'm not a Fortune 500 operating executive anymore, Peter, because the work it's going to take to reinvent and transform is a heavy, heavy lift. But the rewards are going to be so big. Thanks for all the great work and research you do on the AI to RI newsletter, which becomes the content for this. And most importantly, thank you to our listening audience.
- [29:37]
B
You bet.