Why CEOs Need to Lead AI Strategy

Today on the AI Daily Brief, YCEO's need to take the lead on AI strategy, and before that in the headlines, five coding goes mobile. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Alright, friends, quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Zencoder, robots and pencils and super intelligent. To get an ad free version of the show, go to patreon.com/airdailybrief, or you can subscribe on Apple Podcasts. If you are interested in sponsoring the show, send us a note at [email protected]. One more thing, if you are interested in benchmarking information products, original research. I got some big announcements coming up soon. Go to aidbintel.com and sign up for updates, or if you're willing to spend a little time with me, join the AI tracking panel, which is going to be key to some of those initiatives. Now with that out of the way, let's talk mobile vibe coding. Welcome back to the AI Daily Brief headlines edition, all the daily AI news you need in around 5 minutes. We kick off today with a super cool operator themed update. Replet has launched a new feature designed to streamline the process of vibe coding and pushing mobile apps. Now, it's not that previous vibe coding tools haven't had it possible to vibe code mobile apps, and there have even been some platforms like Riley Brown's vibe code that have specifically focused on it. But when it comes to the biggest vibe coding platforms, there were still a ton of barriers. If the goal was actually launching a commercial app, there would be challenges around configuring payments, auditing security, and of course navigating the app store application process. The default then has been to stay on the web. Replet's new features aim to make all of that much simpler. In addition to specifically designing for mobile, after you've built your application, you can publish to the app store with just a few clicks. The pitches that novice developers can complete the entire process without leaving Replet. In an announcement post, the company wrote, "If you've been sitting on an idea, now's the time to bring it to life. Your audience, customers, or community are already on mobile. Your app should be too. From idea to app store in minutes, all on Replet." Perhaps unsurprisingly then, in addition, Bloomberg reports that Replet is closing a new fundraising deal that would see the startup value at $9 billion. Sources said that the round size is around $400 million. Now as exciting as this idea is in theory, it would be easy to not do that well in practice, but the first reports are really good. Eric, who admittedly does work with Replet, said it was tough to keep this one a secret, but Replet now lets you build mode of apps natively, but that's not the exciting part. You can push them directly into the app store with just a few clicks. I've been beta testing this since December, and let me tell you, this changes everything. Collet writes, "Having Replet on my phone while sitting at a coffee shop having a conversation while building my apps is a magical superpower. I'm still in awe every time I use it." Mark Mathson writes, "Okay, I was just invited to test flight a newly created Replet mobile app, published through their new platform feature, and the app was a 10 out of 10 quality all around. Get ready to see a huge increase in quality apps in the app store." Next up, some more funding news for another AI unicorn? Higgsfield has closed a new round of funding at a $1.3 billion valuation. The video Jen starred up said that this was an extension of their 50 million series A, which closed in September, adding a further 80-in-fresh capital. Now, Higgsfield, if you don't know, is a front-end for content creation, serving various open-source video models. The company has been extremely adept, some might even say aggressive at social marketing, with the brand splashed all over X over the last year. Those tactics have paid off, though, with the nine-month-old startup now boasting 15 million users. They said that they've now reached 200 million in ARR, doubling their run rate from 100 to 200 million over the past two months. Higgsfield noted in a press release that this early phase of hypergrowth makes them the fastest startup to 200 million, outpacing lovable cursor open AI slack and zoom. Super.com founder Henry Shee, who now works on AI at Anthropic, and who manages the lean AI leaderboard, confirmed this and said it's basically unprecedented. Now, interestingly, Higgsfield says that 85% of their usage now comes from social media managers. In their words, a major sign that the platform adoption has evolved beyond casual content creation. They also added that adoption is accelerating fastest, quote, "among marketers treating generative video as production infrastructure, running end-to-end workflows," ID8, storyboard, animate, edit, and publish, inside a single system. Next up in the headlines, rumors of more departures swirl as Miramarati's thinking machine's labs faces a full-on talent exodus. On Wednesday, we learned that co-founders Barrett Zulf and Luke Metz, along with Sam Schoenholtz, were leaving TML to rejoin open AI. Together with Andrew Tullock returning to meta in October, that means that TML has now lost three of its six co-founders in a matter of months. Alex Heath of sources now reports that more employees are heading to the exits as well. Heath writes, "Sources say at least a couple others have already resigned from thinking machines after a tense all-hands meeting Miramarati held on Wednesday about Zulf's departure, and more are expected to follow suit. Talks are fluid and it's unclear exactly how many members of Miramarati's small startup will ultimately decamp to open AI." Now, for some, this is just part and parcel of high-stakes Silicon Valley startup building. Tech commentator Robert Scobal wrote, "It's long known in Silicon Valley that if you're a rock star, you usually take a whole team with you. That seems to be what's going on here. What a plunder." Additional reporting from Maxwell's Zephyt Wired suggests that the departures aren't just about people following Zephyt's off-out-the-door. A source that the company said, "This has been part of a long discussion at thinking machines. There were discussions and misalignment on what the company wanted to build. It was about the product, the technology, and the future." Zephyt added, "In the aftermath of these events, we've been hearing from several researchers at leading AI labs who say they are exhausted by the constant drama in their industry. Not so much the denizens of AI Twitter, who are clearly just tuned in for the next chapter of the AI soap opera." Racer X tweeted, "I bet Mira is now also considering going back to open AI." Lastly today, some interesting comments from Google DeepMind CEO, Demis Hassabis, who's warned that Chinese AI models are rapidly closing the gap with their US counterparts. In an interview with CNBC, Demis said that the difference is much smaller than it was a year or two ago, adding, "Maybe there only a matter of months behind at this point." Now at this stage, it's certainly no longer a shock when a highly capable model comes out of China. Zai, Kimi K2, Quen3 are all in the same ballpark as the best models from the West. And in video, Kling is arguably leading the field with their new motion control technology. Still, Demis argued that we haven't seen Chinese labs prove their ability to make truly novel breakthroughs. He said, "The question is, can they innovate something new beyond the frontier? I think they've shown they can catch up and be very close to the frontier, but can they actually innovate something new like a new transformer that gets beyond the frontier? I don't think that's been shown yet." Continuing on the theme of innovation, he said, "To invent something is about a hundred times harder than it is to copy it. That's the next frontier, really, and I haven't seen evidence of that yet, but it's very difficult." The key point, which will be well known to all of you, is that China is no longer dissently behind when it comes to AI. Jensen Huang recently commented that they're actually ahead in some aspect, saying, "China is well ahead of us on energy. We are way ahead on chips. They're right there on infrastructure, and they're right there on AI models, and the areas where the US leads are no longer guaranteed." Earlier this week, ZAI unveiled their new first model trained entirely on Huawei chips software. The model was a relatively small image model, so this wasn't a frontier LLM training run, but the announcement was a proof of concept that Huawei now has a fully capable AI development stack. I think Chinese models are going to do nothing but grow an importance this year, and how that impacts the AI race will have to see. For now that's going to do it for today's headlines, next up, the main episode. Hello, friends. If you've been enjoying what we've been discussing on the show, you'll want to check out another podcast that I've had the privilege to host, which is called "You Can With AI" from KPMG. Season 1 was designed to be a set of real stories from real leaders, making AI work in their organizations, and now season 2 is coming, and we're back with even bigger conversations. This show is entirely focused on what it's like to actually drive AI change inside your enterprise, and as case studies, expert panels, and a lot more practical goodness that I hope will be extremely valuable for you as the listener. Search "You Can With AI" on Apple Spotify or YouTube and subscribe today. If you're using AI to code, ask yourself, "Are you building software, or are you just playing Prompt roulette?" We know that unstructured prompting works at first, but eventually it leads to AI slap and technical debt. Enter Zenflow Zenflow takes you from vibe coding to AI first engineering. It's the first AI orchestration layer that brings discipline to the chaos. It transforms freeform prompting into spec-driven workflows and multi-agent verification, where agents actually cross-check each other to prevent drift. You can even command a fleet of parallel agents to implement features and fix bugs simultaneously. We've seen teams accelerate delivery to x to 10x. Stop gambling with prompts. Enter orchestrating your AI, turn raw speed into reliable production-grade output at Zenflow.free. Most companies don't struggle with ideas. They struggle with turning them into real AI systems that deliver value. Robots and pencils is a company built to close that gap. They design and deliver intelligent cloud-native systems powered by generative and agentic AI, with focus, speed, and clear outcomes. Robots and pencils work in small, high-impact pods, engineers, strategist designers, and applied AI specialists working together to move from idea to production without unnecessary friction. Powered by RoboWorks, their agentic acceleration platform, teams deliver meaningful results including initial launches in as little as 45 days depending on scope. If your organization is ready to move faster, reduce complexity, and turn AI ambition into real results, robots and pencils is built for that moment. Start the conversation at robotsandpensals.com/aidlybrief. That's robotsandpensals.com/aidlybrief. Robots and pencils, impact at velocity. Today's episode is brought to you by Super Intelligent. Super Intelligent is a platform that very simply put is all about helping your company figure out how to use AI better. We deploy voice agents to interview people across your company, combine that with proprietary intelligence about what's working for other companies, and give you a set of recommendations around use cases, change management initiatives, that add up to an AI roadmap that can help you get value out of AI for your company. But now we want to empower the folks inside your team who are responsible for that transformation with an even more direct platform. Our forthcoming AI strategy compass tool is ready to start to be tested. This is a power tool for anyone who is responsible for AI adoption or AI transformation inside their companies. It's going to allow you to do a lot of the things that we do at Super Intelligent, but in a much more automated, self-managed way, and with a totally different cost structure. If you are interested in checking it out, go to AIDailyBrief.ai/Cumpus, fill out the form and we will be in touch soon. Welcome back to the AIDailyBrief. Today we are looking at the results of a couple of recent enterprise AI surveys. Now, if you are a regular listener, you will know that I love it when we get actual data, especially longitudinal data that gives us a sense about not only where things are but how they're changing over time. And one of the big things that stands out in both of these surveys is a shift in and around the leadership of AI initiatives. Simply put, it appears to increasingly be the case that CEOs are no longer delegating AI. And frankly, I think that this couldn't come soon enough. The type of challenges that AI is going to represent going forward are much bigger than the types of challenges that they've represented so far. Now the opportunity is commensurately large, but there is simply no way to get around the fact that this needs leadership from the absolute top of the organization. So first let's look at these studies and then we can get a little bit deeper into the analysis. The first study that we're looking at is KPMG's most recent quarterly pulse survey. The pulse survey looks at leaders of large organizations with a billion or more in revenue and gives us one of our best chances to see how attitudes have changed over time. And the big story here is not just that AI is no longer experimental. Instead, it is that it is absolutely fundamental to the way that organizations see their future. And increasingly, a cleaving differentiator between enterprises who are leading in those who are lagging. As KPMG's global head of AI and digital innovation Steve Chase put it, AI isn't just an investment, it's becoming the backbone of enterprise strategy. What the numbers don't show is the growing divide. While some organizations stall after early deployments, the leaders are scaling fast and pulling ahead. For those treating AI as a true disruptor, this isn't about catching the next wave. It's about agents fundamentally changing how value is created and sustained across the enterprise. And one of the big stories going along with this theme of CEO leadership is as KPMG puts it, AI investment becoming recession proof. First of all, organizations are planning substantial investments, on average saying that they plan on spending $124 million on AI over the next 12 months. That's up from Q1's 114, up from the Q2 dip of 88, and near Q3's peak of 130 million. 67% of the leaders surveyed said that AI will remain a top investment priority for their organization even if a recession occurs in the next 12 months. 59% say that they will continue to invest in AI even if they cannot measure tangible ROI. And this I think has been a key theme of AI since the beginning. Unlike many new technologies, AI is so transparently powerful that there's never really been a question about whether organizations were going to invest in it. The long-termism that you see in that 59% statistic is reflective of the fact that leaders understand that this is something that they have to invest in over time and that they will get benefits from. However, that is not to say that they are not optimistic about being able to measure tangible ROI even in the short term. One of the things that was interesting going back to a different KPMG study from the end of last year, their 2025 global CEO Outlook, was first a similar contrast between enthusiasm in AI and concern about global volatility, but also a major pull forward in their expectations of seeing value return from their AI investments. In 2024, the average CEO in KPMG survey anticipated that they'd see ROI from their AI investment in three to five years, 63% of them nearly two thirds thought that it would take that long. 16% said it would take more than five years, and about a fifth of the CEOs were more optimistic saying it would take one to three years. In their 2025 survey that completely shifted. Now 2/3 said that it would take one to three years, a fifth said it would take just six months to a year, and only two percent said that it would take more than five years. So did those numbers show up in the Q4 pull survey as well? The short answer is yes. 59% of respondents said that they expect to have measurable ROI in the next 12 months. Now alongside this, what people are actually measuring is diversifying and expanding as well. While productivity, profitability, and revenue generated remain top ROI metrics, the way that AI improves decision making among the C suite has jumped nearly 20 percentage points up as a measure of ROI as well. This is something we saw in our AI ROI benchmarking survey as well, where organizations were increasingly focused not just on first order impacts like time savings and increasing output, but more strategic value like improved decision making. In fact, in our survey, organizations deploying use cases that were focused on improving those strategic areas like decision making were reporting a higher ROI overall. Next up, let's talk about agents. The way the KPMG sums things up is that leaders are professionalizing agents, and I think that's a fairly accurate summary. A slightly less unreservedly positive spin, but one that I think still amounts to a good thing is that it feels to me like a lot of these numbers around agents show a clear maturation of the organization's understanding of what the hell agents actually are and what it means to implement them. Get out of the gate. One thing we should note is that the percentage reporting AI agent deployment is actually down a fair bit from Q3. Looking at KPMG's poll surveys from last year, in Q1, 11% of organizations reported deploying agents. Now keep in mind that's not experimenting or piloting that's actual deployment. In Q2, the number was 33% and Q3, the number was 42%. In this poll survey, the number was 26%. And I think that there are a couple different interpretations. One is to say, hey, I guess something happened and deployments are actually down. And that's, of course, totally possible. A second is basically to ignore the intermediate numbers and focus on the fact that year over year, you're still talking about more than a doubling of AI deployments. Certainly, a reasonable take in one that KPMG shares here, although they're certainly not trying to bury the fact that there was a shift down from Q3. I think what we're dealing with again is a maturation and in respondents understanding of what agents actually are. The 42% number was extremely high, and my guess reflected, a, enthusiasm and excitement about agents, b, the fact that there is some distance between this set of survey respondents and the people who are actually on the ground doing these deployments, and c, a little bit of terminology confusion and blending between automations and agents, which is patting the numbers. Not that the ROI survey that we did should be seen as definitive in any way, but just to provide a comparative look, we didn't ask people to categorize their use cases as assisted automated or agent, we did that analysis on the backside ourselves. And what we found is of the over 5,000 use cases that people shared their ROI for, around 56% were assisted, about 30 or 31% were what we would consider automations. And just a little under 14% is what we would consider truly agent at AI that really achieves that level of autonomy that it can be considered as such. Again, I don't want to read too much into this, but if you add up the automation and agent pick number, it's pretty close to that 42% number that KPMG saw last quarter. My guess then is that this 26% number is in fact a lot more accurate when it comes to the percentage of organizations actually deploying agent AI. And I think some other notes from the survey support that maturation thesis as well. About 2/3 of leaders cite agentic system complexity as the top barrier, and a lot of the challenges of using agents show up in these numbers as well. Unconsistent use of agents across the organization was up to 45% from 19% and Q2. Unclear enterprise strategies were up from 20% to 32%, 41% cited a lack of organizational infrastructure, which more than tripled in the last two quarters as a challenge when it comes to agentic AI. That's basically exactly what you would expect to see if you're watching enterprises closely. In fact, that 41% number is going to do nothing but rock it up as people actually run up against the barriers of what agents can do without complete data and context strategies. Now, interestingly, there's also a lot more understanding around how agents are impacting the workforce. 41% of respondents said that agents had impacted how they hire for experienced roles, and for entry level hires that number was at 64%. Entire new roles were emerging. 71% have hired some sort of AI prompt engineer, 59% have hired an AI performance analyst, 58% have hired an AI trainer or data curator. 76% of respondents would pay up to 10% more for candidates who demonstrate strong AI skills, and in addition to just knowing how to use AI, the skills that surround usage of AI are also more in demand. 63% said that they're looking for adaptability and continuous learning in their entry level hires, with 61% saying they're looking for critical thinking and problem solving. Now, the deeper that organizations get into these more complex deployments, the more their challenges start to reflect that as well, specifically we're seeing much more focus on cyber security and how to mitigate that risk. Half of the leaders surveyed plan to allocate between 10 and 50 million in the coming year, two hard and model governance, improved data lineage, and secure agent architectures. 71% say cyber is a key factor for re-evaluating Gen/AI strategy, and a full 80% say that one of the greatest barriers to meeting the goals of their AI strategy is cyber security. For KPMG, this is all pointing in the direction of moving away from isolated agent deployments towards complete orchestrated agent ecosystems. But let's come back to this idea of CEOs taking charge. One of the interesting things that we found in our AI ROI study was that people who reported being sea levels or founders reported significantly higher ROI than other title levels. The mean ROI for use cases from sea levels and founders was 3.59, that means between three, which is modest, and four, which is significant. In fact, more than 50% of use cases overall from those sea levels and founders had high ROI impact categorized as significant or transformational by almost double sea levels and founders reported more transformational use cases, and they were the least likely to report negative ROI as well. Now, there's a bunch of reasons why we thought that might be. One is attribution clarity, i.e. being able to see the entirety of a process play out. Another has to do with correlation where the types of use cases that sea levels are focused on are the ones that inherently would be more transformational. What's interesting though is that another corporate AI survey, this time the BCG AI radar, shows that the sea level is getting much more involved when it comes to AI in 2026. Now in addition to seeing a very similar phenomenon as KPMG around continuing to invest in AI, even if it doesn't show tangible ROI, in fact, in this survey, 94% of CEOs said that they'll continue to invest even if it doesn't pay off in 26. One of the big takeaways as they put it is that AI corporate transformation is moving from a CIO led to a CEO led initiative. 72% of CEOs said that they are the main decision maker on AI in their organization. Like the KPMG CEO survey, optimism on ROI has been pulled forward. 82% of CEOs said that they're more optimistic about AI's potential for ROI this year than they were last year. Now, part of the reason that CEOs are getting involved is that they see it as existential for themselves. Half of the CEO survey to believe that their job stability depends on getting AI right. Overall, CEOs show stronger conviction in AI than other executives. They are the most confident AI will pay off. They most expect major role disruption in the organization over the next five years, and they are most ready to lead in AI transformation. One interesting divide, which again reflects things that we've seen basically ever since CHATGBT launched, is that there are regional differences. CEOs in the West and Europe in the United States seem to be acting because they fear falling behind, as opposed to CEOs in China and India, who are more likely to act on AI because they see value. Still overall, CEOs are enthusiastic. When it comes to agents around 90% of CEOs believe that agents will enable their organizations to report measurable ROI this year, and the amount of overall AI spend that they've committed to Agendica AI for this year is above 30% now. To me, what's maybe even more interesting than just some of these enthusiasm numbers is how it's clear that CEOs see just how much change is actually going to happen because of AI. 58% of leading organizations in this study expect to change in governance and decision rights due to AI, and 90% of CEOs believe that by 2028 AI will significantly transform what success looks like. On this weekend's Long Read Sunday show, I started to get into what I think is one of the major themes beginning 2026, which is a new inflection point, which is fundamentally changing the disruption profile of AI for big organizations. If I'm right about that shift, it is completely necessary that CEOs take the lead, because they're the only ones who have enough power to actually make the changes that are going to be necessary. Still, super interesting to see so much consistency across multiple studies. For now, that's going to do it for today's AI Daily Brief. Appreciate you listening or watching as always, and until next time, peace. [MUSIC]