The AI Daily Brief: Artificial Intelligence News and Analysis - AI Agent Deployments Quadruple in 2025

Starting point is 00:00:00 Today on the AI Daily Brief, Enterprise Agent Deployments quadruple in 2025. Before that in the headlines, why Microsoft's CEO is haunted by the prospect of the company not surviving the AI era. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. All right, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, Notion, Robots and Pencils, Blitzy, and Agency.org. To get an ad-free version of the show, go to patreon.com. slash AI Daily Brief. If you're interested in sponsoring the show, shoot us a note at sponsors at AIdailybreep.a.i. And a reminder that if you are interested in helping us manage the overflow of

Starting point is 00:00:42 customers we have for our agent readiness audits over at Super.com. Send an email to jobs at B-super.ai with a video of you sharing some automation you've created. And with that, let's dive in. Welcome back to the AI Daily Brief Headlines edition, all the daily AI news you need in around five minutes. Today we start off with something that's a little bit different because it's actually the least newsy of our headline stories. There's a bunch of stuff that is theoretically bigger news, but it's a really interesting moment that I think tells the meta story of what's happening right now in ways that we don't always get to see. The history of tech is, of course, littered with names that failed to make the transition as the landscape shifted around them. Xerox, Bell Labs, IBM,

Starting point is 00:01:20 each were a dominant tech company of their era just as Microsoft is today. It turns out that Microsoft's CEO, Satchinadella, is, as the Verge puts it, haunted at the prospect of Microsoft not surviving. Speaking to a company-wide town hall last week, Sotia said, some of the biggest businesses we've built might not be as relevant going forward. Our industry is full of case studies of companies that were great ones that just disappeared. I'm haunted by one particular one called DEC. Now, DEC was a mini-computer manufacturing company in the early 1970s. The first computer Nadella used was a DEC, and growing up, all he wanted to do was go and work for the company.

Starting point is 00:01:56 However, by the early 90s, they had been thoroughly out-competed by IBM after making a string of failed bets on emerging architecture. In fact, Nadella commented, some of the people who contributed to Windows antique came from a DEC lab that was laid off. I think about that and I think about what it takes for a company not just to thrive at one time, but to continue to actually have the smartest best people. The comments came in response to an employee in the UK who noted that Microsoft felt, quote, markedly different, colder, more rigid, and lacking in the empathy we have come to value. And what he might be referring to is the fact that while the entire tech sector has seen layoffs over the past year, Microsoft has had a number of waves where they slashed their headcount.

Starting point is 00:02:34 Indeed, Tom Warren of the Verge wrote, I've spoken to dozens of employees over the past few months, and they all told me that morale inside Microsoft is at an all-time low. Satya continued, here we are at our 51st year as a company, and if you look at a set of metrics, we are thriving. But at the same time, when I think about the degree of difficulty that is ahead, for us to navigate what is a changing industry, a changing tech sector, and changing economics, we have some very hard work ahead of us. He said, pointed, all of the categories that we may have even loved for 40 years may not matter. Us as a company, us as leaders, knowing that we are really only going to be valuable going forward if we

Starting point is 00:03:07 build what's secular in terms of expectation instead of being in love with what we've built in the past. At a time of platform shifts, you want to make sure you lean into even the new design wins, and you don't just keep doing the stuff that you did in the previous generation. You would rather win the new than just protect the past. Look, here's what I would say. At risk of this going long into a full main episode length type of episode, Microsoft came out of the gate very impressively when it came to the Gen.A.I. Because of their unique partnership with OpenAI, which not only set a template for partnerships of this era, but also positioned their products as early leaders in actual capabilities.

Starting point is 00:03:42 And yet from there, in my estimation, it has been stumble after stumble. I don't think they've done a good enough job, keeping pacing between the quality of their co-pilot products and what people can sign up with their g-mails, which is one of the biggest complaints that we hear constantly from employees at enterprises, and one of the biggest culprits and why they're still such an epidemic of shadow AI. I think their response to the firing and rehiring of Sam Altman, while I am sympathetic, that they had to, from a fiduciary responsibility perspective, start to hedge a little bit and not put all of their chickens in that seemingly chaotic basket. The hiring of a person who's decided to focus their AI efforts on consumer,

Starting point is 00:04:15 rather than just aggressively taking advantage of the incredible distribution power they have in their enterprise business, just seems insane to me. Now, that distribution power all on its own, and the amount of lock-in that they still have with businesses means that I am absolutely not counting them out. I think the moves they've made recently, for example, being willing to put Claude into GitHub co-pilot because of what they perceive as its better quality is the type of thing they're going to have to do to keep on moving.

Starting point is 00:04:40 But I will say, whereas, for example, the prospects of Google look radically better than they did two years ago when it comes to AI, I think it's pretty hard to deny that the prospects of Microsoft look anything but worse right now. That said, it is exactly this, sort of haunting that Satya Nadella is talking about at these meetings that can lead to the big decisions that help companies avoid that fate, and I wish them nothing but luck, and so I am excited

Starting point is 00:05:03 to see what big moves they might pursue. One thing that could be really valuable is while people gripe and moan, they still use teams, and Microsoft is now adding agents throughout their team's platform. Agents will now sit in and take notes in meeting, help create schedules, and lurk in every channel waiting to help. The basic idea is to have an AI meeting note taker that's also integrated into corporate knowledge bases and other parts of the Microsoft stack, and while that agentic utility may seem simple or basic, it could be extremely valuable for a significant number of teams. The company is also making big bets on infrastructure. Microsoft also announced that they are in the final stages of construction on a $3.3 billion

Starting point is 00:05:38 project in Wisconsin. They plan to break ground on an additional $4 billion facility in the area shortly. The data centers are built on land purchased from the failed Foxcom plant, which was commissioned to manufacture LCD screens in 2017 but was never completed. Microsoft is benefiting from building on top of the partially installed infrastructure and site preparation. President Brad Smith says the facility will house hundreds of thousands of Nvidia Blackwells, and the facility is being designed as a gigantic new training cluster capable of developing Microsoft's next generation of in-house models.

Starting point is 00:06:06 Last week, Microsoft's AI CEO Mustafa Suleiman said that the company is making significant investments in the compute capacity they would need to train frontier models for themselves, and this seems to be the manifestation of those plans. Moving over to Microsoft competitor Google, that company is adding a new actor into the AI browser wars. The TLDR is that Gemini is going to be baked in as a default feature of Chrome. Google is also adding AI mode to the Chrome search bar, making it easier to access. While right now there are some very mild search and information-related agentic features

Starting point is 00:06:36 over the coming months, Google plans to introduce many more. They appear to be starting with the standard fare for web agents, things like automatically booking flights and shopping for groceries, but have plans to build in more cross-platform agentic tasks, like being able to sync up across Google Calendar and Workspaces. The sheer scale of their distribution is such that for many people, a random little button that says Gemini or AI mode in Chrome, is likely to be one of their early interactions with this entire space. Now, if you're noticing a theme in everything getting quietly agentified, you are not

Starting point is 00:07:06 wrong. Notion has also revamped their platform around agents calling it Notion 3.0. The way that the company describes it is this. They write anything you can do in Notion your agent can do too. The busy work that fills your day can now be done in minutes. The new Notion agents can create pages and databases, automatically update data, and complete dozens of other tasks. Notion said that their agents can complete up to 20 minutes of work across hundreds of pages

Starting point is 00:07:30 at once. They can also connect with outside data sources, including Slack, email clients, and Google Drive. I haven't had a chance to use it yet, but the first reports on extra positive. Lily Bodner writes, been using the new AI agent from Notion for the past couple of weeks, and it's awesome, super fun to use. normally have multiple agents going across different tabs. Set up a Notion's rule file and created a bunch of new databases that I've been putting off doing manually.

Starting point is 00:07:52 As we round the corner on this year and start to think about what the story of 2026 is going to be, the agentification of everything and frankly the normalization of agents is, I think, positioning itself to be one of the big themes. It's certainly something that we will continue to keep an eye on here, but for now, that is going to do it for the headlines. Up next, the main episode. Chatbots are great, but they can only take you so far. I've recently been testing Notion's new AI agents and they are a very different

Starting point is 00:08:19 type of experience. These are agents that actually complete entire workflows for you in your style and best of all, they work in a channel that you already know and love because they are purpose-built Notion super users. Notion's new AI agents completely expands the range of what Notion can do. It can now build documents from your entire company's knowledge base, organize scattered information into organized reports, basically do tasks that used to take days and get them incomplete in minutes. These agents don't just help with work, they finish it. Getting started with building on Notion is easier than ever. Notion agents are now your very own super user to help you onboard in minutes. Your AI teammates are ready to work. Try Notion AI for free at the link in our

Starting point is 00:08:57 show notes. Small, nimble teams beat bloated consulting every time. Robots and pencils partners with organizations on intelligent cloud-native systems powered by AI. They cover human needs, design AI solutions, and cut through complexity to deliver meaningful impact without the layers of of bureaucracy. As an AWS-certified partner, robots and pencils combines the reach of a large firm with the focus of a trusted partner. With teams across the U.S., Canada, Europe, and Latin America, clients gain local expertise and global scale. As AI evolves, they ensure you keep peace with change, and that means faster results, measurable outcomes, and a partnership built to last. The right partner makes progress inevitable. Partner with robots and pencils at robots and

Starting point is 00:09:39 Pencils.com slash AI Daily Brief. This episode is brought to you by Blitzy, the Enterprise Autonomous Software Development Platform with Infinite Code Context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale code bases with millions of lines of code. Enterprise engineering leaders start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and pre-compiles code for each task.

Starting point is 00:10:07 Blitzy delivers 80% plus of the development work autonomous. while providing a guide for the final 20% of human development work required to complete the sprint. Public companies are achieving a 5x engineering velocity increase when incorporating Blitzie as their pre-IDE development tool, pairing it with their coding co-pilot of choice to bring an AI-native STLC into their org. Blitzy is providing a limited time, 30-day free proof of concept for qualifying enterprises. The team will provide a 5x velocity increase on a real development project in your org. Visit blitzy.com and press book demo to learn how Blitzie transforms your STLC

Starting point is 00:10:39 from AI assisted to AI native. That's BLITZY.com. Shape the future of Enterprise AI with Agency, AGNTCY. Now in Open Source Linux Foundation Project, agency is leading the way in establishing trusted identity and access management for the Internet of agents, the collaboration layer that ensures AI agents can securely discover, connect, and work across any framework. With agency, your organization gains open, standardized tools, and seamless integration, including robust identity management to be able to identify, authenticate, and interact across any platform. Empowering you to deploy multi-agent systems with confidence, join industry leaders

Starting point is 00:11:17 like Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and 75-plus supporting companies to set the standard for secure, scalable AI infrastructure. Is your enterprise ready for the future of Agentic AI? Visit agency.org to explore use cases now. That's agn-tcY.org. Welcome back to the AI Daily Brief. Today we are looking at the latest KPMG AI quarterly poll survey. And this really is something of an enterprise AI report card. This is now the fourth edition of this study since the end of last year. And given how fast things are moving, the changes quarter over quarter really do provide an interesting chance to see how broader technology change is translating back into organizational process and

Starting point is 00:11:59 expectation. The survey methodology involves conversations with leaders from 130 organizations that have at least a billion dollars in revenue. So this represents large companies. KPMG's headline for the press release was agent deployment accelerates as organizations build confidence through early wins. And I think that there are three very clear and interesting stories here. One is, as they captured, agent deployment growing. In fact, agent deployments have quadrupled since the beginning of the year. Number two, the normalization of tensions within the workforce. And number three, a new vision for ROI. Now, we're going to get into all of that, but first, I wanted to go back and look a little bit and ground us in where we were at each

Starting point is 00:12:38 these studies. I'm not going to go through them comprehensively, but I do think it's useful to get a sort of longitudinal sense of where things were. If you go back to the Q4-2020 study, it is all about forward-looking and what's coming next. There is this sense that agents are on the horizon, but they are very much still out ahead of us. A lot of the questions are anchored not in what people are doing or what results they've seen, but what they believe AI's impact will be, which frankly, and no disrespect to KPMG, is the sort of filler question that you ask when organizations aren't doing enough to just ask about their behavior. This is where a lot of these types of studies were at the end of last year. Which isn't to say there was nothing about actual actions. When asked

Starting point is 00:13:16 where they were with agents, most of the organizations, a majority, were in the exploration phase. Now, exploration in this context, comes even before piloting and can mean basically anything around trying to understand what agents are going to do for the organization. 37% said that they were piloting AI agents, and 12% said that they deployed AI agents, although I remember strongly thinking even back then that I would really like to know what those 12% called or considered agent deployments because that number seemed a little robust to me. When it came to what they thought agents were going to be used for over the next year, 60% said administrative duties, 54% said call center tasks, 53% said developed new business materials. So a lot of the stuff

Starting point is 00:13:58 that people were now using AI assistance to do better themselves, they were basically imagining agents offloading those things and adding a new layer of autonomy. Now, one really important finding, which I think has lurked around a lot of this year, was that it was almost the inverse relationship of how you would expect technology to permeate across the organization in that it was not bottoms up adoption, but leader-led adoption. When asked if they were using Gen A.I. Tools at their organization, only 15% of entry levels said that they were and 26% of middle managers, but by the time you got up to executive management, it was 51%, and when you got to the C-suite, it was 57%.

Starting point is 00:14:35 This is a pattern that we've seen over and over again, where there is actually a gap in adoption between leadership and employees. Or at least, there is a gap in formal adoption of tools that the organization has approved. The other thing that studies find constantly is that there is actually a huge amount of shadow AI happening that people remain even to this day, although this might be changing a little bit, worried to tell their colleagues and bosses that they are using. Sometimes that's because they think it will delegitimate their work. Sometimes it's because they know they're violating policy

Starting point is 00:15:05 because they're using a tool with their, for example, personal Gmail address that is not approved by their organization but is much better than the version that their organization has approved. So this was the state heading into the year. Between Q4 and Q1, we saw a big jump in trying to make all of this exploration real. The banner headline was an almost doubling of organizations that were piloting agents from 37% in Q4. to 65% in Q1. What's more, the intention was completely ubiquitous. Ninety-nine percent of

Starting point is 00:15:36 organizations said that they planned to deploy AI agents. And interestingly, there was a real strong bias, more than two-thirds to one-thirds towards buying a pre-built agent versus some combination of buying and building. This was the study that also showed assistant category AI becoming table stakes. The percentage of knowledge workers who are using AI productivity tools, i.e. things like Chatchip-T and co-pilot on a daily basis, jumped massively from 20, 2% to 58% in just a single quarter. So we have agent pilots up, assistance becoming table stakes, and we had a big growth in expected investment. The amount that these organizations anticipated spending on Gen.A.I. over the next year jumped from 89 million to 114 million. By the time we get to

Starting point is 00:16:18 quarter two, we are continuing to see rapid change. By far, the biggest headline here was that agent deployments tripled between Q1 and Q2, whereas 11% were in. in the deployment phase back in Q1. By Q2, that was 33%. Commensurately, piloting agents had gone down because so many more organizations had moved past pilots into deployment, but overall, 90% of organizations in that Q2 study, were past the experimentation phase, i.e., they were actively piloting or deploying agents. One of the questions that I thought was most interesting, given how much I talk about the idea of efficiency AI versus opportunity AI, or in other words, a mindset of using AI only to do the things that you do now a little bit faster, a little bit cheaper,

Starting point is 00:17:02 a little bit better, versus doing things that weren't possible before. When asked whether they were primarily concerned with productivity and efficiency, or on the other end of the spectrum revenue growth, nearly half said that they were equally focused on both, 46%. And in fact, it's not in this chart, but there was no one who said that they were only focused on efficiency. Everyone was either focused on revenue growth or some hybridization, which candidly, I think, is as much aspirational as it is descriptive, but I'm still glad to see that that's where the aspiration

Starting point is 00:17:29 is. KPMG vice chair of AI and digital innovation, Steve Chase said, the data shows just how quickly AI agents are moving out of pilots and into production, and that momentum will only accelerate, and that is, of course, the anchor story of the new Q3 pulse survey. The banner standout headline is that agent deployments have this year nearly quadrupled, from 11% back in Q2 to 42% in Q3. Again, this means deploying agents that have moved all the way through a pilot phase into now just becoming part of the organization's operations. Alongside more broad deployments, the challenges are starting to change consequently as well. KPMG writes, the complexity of agentic systems has emerged as a dominant hurdle, jumping from 39% to 71% as organizations

Starting point is 00:18:15 grapple with the intricacies of deploying agents at scale. Now, the second standout statistic that I mentioned from this survey is a potential shift in the relationship between workers and agents. While KPMG had found that 47% of employees were somewhat resistant to agents back in Q2, that number was down to 21% by Q3. We don't have the answer, but the question of why here is, of course, hugely important. Is that because these organizations have done more work to get their teams on board? That's totally possible. One of the trends that we've seen was the gap between leadership and employees when it comes to their understanding of AI initiatives, and there have been a lot of people beating the drum that that needs to be changed before real agendic

Starting point is 00:18:53 adoption can happen. Perhaps this is also just a consequence of the fact that more employees are actually using agents in deployment to make their work lives better. They're discovering that at this stage, at least, agents are not taking their job, but taking chunks of their job that they never liked anyway. All of those are speculations, but that number is a massive, massive shift, and it will be quite interesting to see if that continues in future iterations of the study. The third number that I thought was really interesting in the Q3 survey was this one. 78% of leaders now say that traditional metrics fail to capture the business impact of generative AI. In their announcement post, KPMG wrote,

Starting point is 00:19:31 An overwhelming 78% of leaders now acknowledge that traditional business metrics do not capture AI's full impact, a recognition that speaks to the technology's transformative nature beyond simple cost savings. This evolution in thinking comes at a critical moment. The same percentage of leaders, 78%, report facing sales. significant pressure from investors and boards to demonstrate AI value, creating attention between the need for quick wins and the reality that AI's benefits often transcend traditional ROI calculations. While the majority, 57%, expect measurable ROI within 12 months, which editors note is a big statistic on its own, value from agent investment is already being delivered.

Starting point is 00:20:07 Leaders are tracking improved productivity, 97%, enhanced profitability, 94%, and higher quality work, 91% outputs that can be quantified within traditional ROI frameworks. So basically what they're saying is that leaders are doing two things simultaneously. On the one hand, they are trying to work to fit Gen. AI impact into the traditional ways that we measure and think about ROI for the sake of reporting and organizational understanding. But they are also acknowledging that these metrics ultimately fall short of telling the full story.

Starting point is 00:20:38 This is something that I have heard echoed across every speaking engagement that I've had over the last few months and many, many conversations that we've had at Superintelligent, that the leaders who are deepest into AI feel that the traditional ways that we measure new technologies are just simply reductive. What's more, AI is not monolithic. The way that you measure the impact of assistance is going to be very different than you measure the impact of agents. And by the way, the way that you measure the impact of agents in one department might be very different than you measure it in another department. The impact of coding agents is going to be expressed very differently than the impact of customer service agents. And yet at the same time,

Starting point is 00:21:15 there is such huge conviction of the power and potential of these technologies that what I'm not seeing is any amount of thinking that a lack of ROI is going to suddenly cause a shift in strategy away from Gen A.I. One other thing that I was intrigued to see from the deeper readout of the survey relates to something that I've complained about on here before, which is that I think that organizations aren't doing a very good job around agent-specific upskilling, in large part because the market is not giving them good options for agent-specific upskilling. A lot of the tools and upskilling platforms out there are stuck in a prompt engineering-type paradigm. These organizations are reporting at least their own efforts in this area. For example, 57% said that they're implementing

Starting point is 00:21:53 AI agent shadowing programs where employees observe experts working with agents. 40% said that they're trying to foster a partnership mindset through workshops on human agent-collaboration, and 52% said that they're creating agent-specific sandbox environments where employees can practice interacting with AI agents. By the way, they are not ignoring prompt engineering anymore. The number of organizations that are teaching prompting skills has jumped from 69% to 85%. And if you're looking for a sense of where things are headed in the future, watch what they're spending, not what they're saying. The anticipated investment in Gen. AI has once again jumped from 114 million to now 130 million. So really interesting stuff that basically maps very closely to, I think, what you would expect

Starting point is 00:22:34 just watching the field day to day. Agents are rapidly moving out of experiment. and piloting into production. Employees are getting more used to their new digital colleagues. Thanks, in some cases, it seems, to more concerted upskilling efforts. Leaders are thinking more comprehensively about ROI, and the amount that they're planning to spend just keeps increasing. Let me know what you think about this,

Starting point is 00:22:53 if anything here surprises you or doesn't match your experience, but for now, that is going to do it for today's AI Daily Brief. Appreciate you listening or watching as always, and until next time, peace.

The AI Daily Brief: Artificial Intelligence News and Analysis - AI Agent Deployments Quadruple in 2025

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.