The Great AI Pivot: Labs Refocus on Agents, Enterprise, and the Race for Super Intelligence

[Paul Roetzer]: All the labs realize what Claude Code unlocked. And it wasn't like it was the first coding agent. It was just the best. They did something different with the harness, like how they enabled it to do what it does. All these labs see, not the finish line, but the next mile marker of agentic capability and their ability to automate AI research and their ability to then, as Logan Kilpatrick's deleted tweet said, to start disrupting everything.

Welcome to episode 205 of the Artificial Intelligence Show. I'm your host, Paul Roetzer, along with my co-host, Mike Kaput. We are recording Monday, March 23rd, about 10:00 a.m. Eastern time. Some big stuff last week, Mike. Last week was just crazy. We were on a company retreat for two of the days. So, I always feel like I lost track of time for the week. And then my entire week was spent getting ready for the company retreat. You and I both taught workshops, which we'll talk a little bit about, to the team. And then I did five presentations and workshops, I think, on the first day. So, it was a little bit of a crazy week. But in between all that, we had over 50 different sources in the podcast sandbox this week. As usual, Mike did an amazing job of curating the topics for today, and we were updating what we were going to say even about 3 minutes ago before we came on. We still may adapt it as we move forward. It was just some big stuff—OpenAI and their shift, but it is a larger trend about what's going on with the labs. There's some new polling data about AI. Meta's got a rogue agent. There's a lot to unpack this week.

This week's episode is brought to us by AI Academy by Marketing AI Institute. Talk about AI Academy a lot. This is the core focus of what I do at the company, and it's a huge part of what Mike does at the company is building the content and the curriculum for AI Academy. It's designed to help individuals and businesses accelerate their AI literacy and transformation through personalized learning journeys and an AI-powered learning platform. New educational content is added weekly, so you're always up to date with the latest AI trends and technologies. Our AI for Industries collection features six course series and certificates that are designed to jump-start AI understanding and adoption across industries. The six that are available right now, and they're part of the overall AI Mastery membership program, or you can buy them individually. We have AI for professional services, AI for healthcare, AI for software and technology, AI for insurance, AI for financial services, and the newest one that just came out last Friday was AI for retail and CPG.

These series are an ideal launchpad for organizations that want to level up their teams and accelerate that AI adoption and impact. Mike teaches a number of them, including AI for professional services. Later on in the episode, we're actually going to get some insights from Mike of some of the big takeaways he had from that series. It is probably going to be part of a new element of the podcast. We're going to start trying to drill into some of these course series we're creating. We're spending so much time researching and building these things. We want to bring some of those core insights to everybody as part of this podcast. Mike will tee that off this week with AI for professional services. Individual and business account plans are available now, or you can buy those single courses and series for one-time fees. You can go to academy.smarterx.ai to learn more.

AI Pulse Survey Results

[Paul Roetzer]: All right, Mike. We have our AI pulse this week. You can participate in these pulse surveys each week. They're informal polls of our listeners, where we ask a couple of questions related to that week's episode. Last week, we had Atlassian. Atlassian laid off 1,600 workers and explicitly cited the AI era as the reason. What is your reaction? 39% said this is the new normal; AI-driven restructuring is real and accelerating. 26% said it's AI washing, a fast-growing company using AI as cover for cost-cutting. 25% said it is too early to tell; we need to see if the roles are truly replaced. And then 11% said they are more concerned about the total tech layoffs in 2026 than any single company. Nothing really surprising there. It's a pretty balanced response overall, but 39% is the highest response rate. That is the new normal.

In a New York Times quiz, 54% of readers preferred AI-written prose over human originals. What's your reaction? 39% said not surprised; AI has gotten genuinely good at clean, polished writing. 28% said writing quality was never the real moat; taste, judgment, and point of view are. 20% said this is a wake-up call for professional writers to differentiate beyond service quality. 14% said the quiz was flawed and that it was irrelevant results. We will give you the two pulse questions for this week later on at the end of the episode. But again, smarterx.ai/pulse if you want to participate in those pulse surveys each week.

All right, Mike. So, the first one started in our sandbox as a bunch of OpenAI news. There was a whole lot of stuff. I'm going to let you unpack what happened with OpenAI across the 15 articles that we were looking at. And then I'm going to do my best to take a zoom out and say what is actually going on at all of these labs, because I think there's a major shift happening. When you start looking at the collection of all of this information at the same time, you start to see the trend of where this is going.

AI Labs Refocus on Agents and Enterprise

[Mike Kaput]: Right now, OpenAI is in the midst of executing what might be one of the more dramatic strategic pivots it's done so far. It's simultaneously restructuring how it sells, what it builds, and who builds it, all while preparing for a potential IPO later this year. On the enterprise side, Reuters reports that OpenAI is pursuing partnerships with multiple private equity firms in deals potentially worth a combined $10 billion. These firms include places like TPG, Advent International, Bain Capital, Brookfield Asset Management, and others. The PE investors would contribute approximately $4 billion and receive equity stakes, board seats, and influence over how OpenAI's technology gets deployed across their portfolio companies. The logic here is that private equity firms control massive portfolios of enterprise companies and influence their tech spending. This partnership gives OpenAI a distribution channel directly into those businesses. Notably, Anthropic is also reportedly courting private equity, including Blackstone, signaling that this may become a standard go-to-market playbook for some of these frontier AI companies.

On the product side, OpenAI is consolidating its web browser, Atlas, ChatGPT as a whole, and its Codex coding tool into a single unified desktop, what they call a super app. Fiji Simo, who leads OpenAI's applications division, confirmed this move, saying the company is cutting back on "side quests" to focus on coding and business users. At an all-hands meeting on March 16th, Simo laid out the commercial goal that they want to convert OpenAI's 900 million users into "high compute users" by turning ChatGPT from a consumer chatbot into a productivity instrument built around agentic AI. Interestingly, they're facing quite a bit of competitive pressure on this front. According to enterprise software vendor Ramp, the proportion of businesses using Anthropic increased from 1 in 25 to nearly 1 in 4 within a single year. Anthropic currently wins approximately 70% of direct comparisons against OpenAI in new enterprise contracts.

Meanwhile, at the same time, OpenAI is also going all-in on fully automated AI research. Founding OpenAI member Andrej Karpathy went viral this past week. We talked about this last week, describing an experiment where he deployed an autonomous AI coding agent to run continuous research for 2 days. He calls it AutoResearcher. Basically, this agent executed hundreds of experiments, discovered new optimizations, and sped up how well the model itself worked in terms of its training time. Interestingly, Shopify's CEO tested the same approach on his internal company data, running an agent overnight that conducted dozens of experiments and improved performance by almost 20%. Karpathy says all frontier AI labs will adopt this approach, calling it "the final boss battle" these labs face. There are reports that OpenAI is following suit, going all-in on this idea of trying to build an AI researcher. Lastly, they are reportedly nearly doubling their headcount, according to the Financial Times and Bloomberg, over the next year as they scale across all of these initiatives simultaneously. Paul, maybe connect the dots for me here. OpenAI is making some pretty big, pretty sudden changes.

[Paul Roetzer]: The trend I was referring to goes back to episode 189 of the podcast on January 6th. Right out of the holidays was when Claude Code blew up and it became very hot over those last 2 weeks of 2025. We spent an entire segment of the episode talking about what was happening with Claude Code and how something had definitely changed. That was the starting point, and all the major AI labs are in this accelerating race for autonomous agents and enterprise customers. That's the thing I reference—we first started the outline for this podcast yesterday, and there was just this focus on OpenAI, but when you look in the totality of all the articles we're looking at, all the tweets we're seeing, you see that everything has changed to this refocus on agents and enterprises. This was not really OpenAI's core. It's not like they weren't going after that audience and they weren't building agents before, but Claude Code changed things. You and I can attest to this. It's incredible. Within Claude, the ability to build things—I'll give an actual example a little later on in this episode—it changed things and they're ahead of everyone, very clearly ahead when you use the product.

I'll break down a little bit the OpenAI thing, but then I want to get into the bigger picture. You mentioned Fiji Simo's talk about private equity firms and that they're in these advanced talks, and that both OpenAI and Anthropic are aggressively courting these PE firms. This makes a ton of sense and we've talked about this a little bit before on previous episodes, but Anthropic, as you mentioned, is winning in this space. OpenAI's enterprise business, according to Reuters, is 10 billion out of the total analyzed revenue of about 25 billion right now. That's a run rate; they're not actually at 25 billion yet in a year, but that's the run rate they're on right now.

She tweeted on March 16th, "We're excited to be building a deployment arm and we'll share more details soon." That's what I was talking about—this idea of getting out with these frontier alliances where they're actually working with the consulting firms and stuff. There's just a lot going on where they're trying to get to where the enterprise customers are. When it started getting into this idea of refocusing, which is interesting because I remember last fall we were talking about this—all of a sudden Sam Altman's everywhere. They're going to do space stuff. They're going to do robots again. They're going to build the video gen apps and social networks and devices with Jony Ive. They're just everywhere and it was like, woah, you're getting crushed right now on the model side. Why don't you focus on the model side? It appears they've come to realize that.

Fiji Simo tweeted on March 19th, "Companies go through phases of exploration and phases of refocus. Both are critical, but when new bets start to work like we're seeing now with Codex, which is their version like Claude Code, it's very important to double down on them and avoid distractions. Really glad we're seizing the moment." When I first saw that tweet, I thought that was weird. It's just a weird tone on a tweet, almost like people were questioning whether she was behind this focus because that's not what she was brought there to do. She was in part brought to diversify based on her background. I think some people may have taken this news as almost a slight against what she was supposed to be doing there. That's how I read that tweet. It was a really interesting effort to set the tone that you're behind all this.

That was in relation to the Wall Street Journal article that said OpenAI plans launch of desktop super app to refocus and simplify user experience. In that, there was a quote: "We realized we were spreading our efforts across too many apps and stacks and that we need to simplify our efforts." That was from Simo. "That fragmentation has really been slowing us down and making it harder to hit the quality bar we want." It said top executives including Altman, Chief Research Officer Mark Chen, and Simo have spent the last few weeks reviewing OpenAI's product portfolio and looking at areas to deprioritize. In an all-hands meeting, she told employees they couldn't afford to be distracted by those side quests you mentioned and that they're in this major battle with Anthropic and it's basically a code red internally.

This is all related to this idea of a fully automated researcher, which isn't news and we've talked about this being something they're working on for at least the last year. But I think the timeline is starting to become more clear. They say their new research goal is the North Star for these next few years, pulling together multiple research strands including work on reasoning models, agents, and interpretability—meaning knowing what the models are doing and why they're doing it. There's even a timeline. OpenAI plans to build an autonomous AI research intern, a system that can take a small number of specific research problems by itself by September. A lot of what Andrej Karpathy is talking about is a prelude to this stuff. It said the AI intern will be the precursor to a fully automated multi-agent research system that the company plans to debut in 2028. It's a weird timeline to me. I don't know why it would be that long. But anyway, this AI researcher OpenAI says will be able to tackle problems that are too large or complex for humans to cope with.

You mentioned the idea of these side projects. When you hear side projects, it could be things like the Sora video generation app, like the standalone app. I have to think the planned hardware devices fit into this bucket. If you didn't spend the 6 billion on Jony Ive—but I have to imagine that there's a chance you get delays in the hardware because that's hard. It's a difficult thing to pursue and that could definitely be a major distraction. E-commerce features in ChatGPT, you could see those kind of get sidelined. There are lots of interesting things they've been doing that could get sidelined in all of this. Then you mentioned that in the same time they're doubling headcount. They're aiming to grow to about 8,000 employees. They're at about 4,500 today according to Financial Times.

Overall, it creates this muddy relationship continuing with Microsoft as well. When I started zooming out, I wondered what's going on with all the other labs. We hear so much lately about the challenges Microsoft and OpenAI are having as they try to reimagine that relationship so that OpenAI could get in a position to go public. In the process, they allowed them to start developing partnerships with people like Oracle and AWS, which I'll talk about in a moment. Then we get into the Microsoft thing. We'll talk a little bit more about this one in a rapid fire and we'll drill into this. But the premise is Microsoft made a major shift last week where they're moving Copilot under Satya Nadella. They're actually moving it under another executive, Jacob Andreou, but then he reports directly to Satya and they're taking Mustafa Suleyman, who was in charge of Microsoft AI, and he's going to just run the super intelligence lab, it sounds like. At the same time, Microsoft according to Financial Times is weighing legal action over 50 billion dollar Amazon OpenAI cloud deals. Now, we have this weird muddying of relationships between Amazon and OpenAI.

You have xAI. Elon Musk tweets on March 12th, following lots of turnover at the AI lab. A lot of the co-founders have left in the last 60 days. He tweeted, "xAI was not built right first time around, so is being rebuilt from foundations up. Same thing happened with Tesla." So, you have xAI, one of the major labs, in basically a complete reset mode. This is a month after on February 12th, they got acquired by his other company SpaceX. So, just 40 days ago, SpaceX said on Monday that it acquired xAI, the AI company controlled by Musk, to consolidate his empires and build this one unified company. That combined company now includes X, like the Twitter platform, and includes xAI, and then they have a deep relationship now with Tesla. At the same time, Musk is suing OpenAI, and that's supposed to go to trial in April. You have this crazy thing, but Elon Musk is watching what's happening with agents and enterprise. He wants a piece of that and he realizes they didn't build this the right way. Nobody hits the reset button faster than Musk. If something is not working, he's going to blow it up.

Then you have Meta. March 12th, Meta delays roll out of new AI model after performance concerns. They're spending well over 15 billion last year just on talent acquisition. They're investing heavily. They're rumored to be spending 135 billion this year on CAPEX to build out the future. It doesn't seem to be working yet. They haven't released a major model since they acquired Alexander Wang and Scale AI. Meta's sort of in upheaval. They've kind of fallen off. Them and xAI are just sort of down at the bottom right now. You had Yann LeCun leave xAI, but then Meta shows up and buys Multibook, the AI agent social network that went viral because of fake posts earlier this year. Meta's trying to get in and have a piece of this agent game. They'd probably love to play in the enterprise world, but that's not their natural thing.

Then you have Jensen Huang talking last week about Open Claw being the next ChatGPT. There's a CNBC article that says Jensen Huang, the CEO of Nvidia, on Tuesday pointed to a fast-rising AI project called Open Claw as a major step forward in how people interact with artificial intelligence. He said it is now the largest, most popular, the most successful open-source project in the history of humanity. This is definitely the next ChatGPT. Open Claw is an open-source autonomous agent platform that goes beyond traditional chatbots. Instead of answering questions, these agents can complete tasks, make decisions, and take actions with minimal input from users. Nvidia moved quickly to build around Open Claw's momentum. The chip leader on Monday announced Nemo Claw, an enterprise-grade version of Open Claw that layers Nvidia's software stack and tools on top of the platform.

Then you have Google DeepMind. Google came in hot with Gemini 3. It was great. It's powerful. They've just last week announced some major improvements to Gemini within Google Workspace, which we experience every day. We use Google Workspace and we embed Gemini. They've had kind of a runaway success with Notebook LM. Even though, when you talk to the average business leader, they have no idea what Notebook LM is. In our bubble, Notebook LM is amazing. We talk about it all the time. We have courses on it. The average person has no idea what it is or how to use it. They've had success building these individual apps like Notebook LM and Gemini. They announced a major investment last week in AI Studio, where they're trying to get into the vibe coding game, trying to play along with how Claude Code and stuff is. But the reality is the AI Studio is still for developers. I don't know how to use it. I went in there last week and thought maybe it's ready for me to use it, and it's not.

Gemini, while amazing, Google DeepMind has no answer to Claude Code right now. It's running circles around them. Based on what a Google engineer said—we talked about this on episode 189—Yana Dogan, a principal engineer at Google, on January 2nd tweeted: "I am not joking and this isn't funny. We have been trying to build distributed agent orchestrators, which is exactly what we're talking about with like Open Claw and Claude Code, at Google since last year. There are various options. Not everyone is aligned." I still can't believe this tweet was allowed to go out. "I gave Claude Code a description of the problem. It generated what we built last year in an hour. It wasn't a very detailed prompt and it contained no real details given I cannot share anything proprietary. I was building a toy version on top of some of the existing ideas to evaluate Claude Code. It was a three-paragraph description." And then when will Gemini get to this point? I think someone asked. He said, "We're working hard on it right now, the models and the harness."

I thought this was really interesting, Mike. Logan Kilpatrick, who's sort of head of AI developer relations—he's a major player within Google DeepMind, came from OpenAI—he tweeted, and I couldn't believe I saw this tweet. I thought that was going to come down fast, and it did. He deleted it. It said, I think this was on Saturday, "All the industries you thought weren't going to be disrupted by AI are about to be disrupted." They're not allowed to say that. Google customers are reading that saying, "I'm sorry, what?" It is 100% true, but you can't say that. Someone got that down real fast. Google is sort of in this crazy phase where they're trying to build it into Gemini. They're trying to make it function within the productivity tools that they have while DeepMind is telling you that every industry is going to be changed.

I'll wrap here with what I think is—if you want to, you have to be ready for the technical stuff when you listen to Andrej Karpathy. Mike and I talk about Andrej all the time. He ran Tesla computer vision for 5 years. Co-founder of OpenAI. Did a bounce back to OpenAI for about a year. Now he's an independent researcher. He's been on fire on X the last 3 weeks. Just all these crazy things he's working on. But he did an interview on the No Priors podcast. If you're ready for the technical side of this, listen to this episode. We'll put it in the show notes.

A few key notes that I was taking—I was listening to this yesterday, actually. He was talking about how fast these models have evolved and how it's largely a skill issue, which is funny because that's a term my son tells me when he beats me in a video game. He says, "It's a skill issue, Dad." If I lose in Mario Kart, I say, "Oh, it's the wrong character." He goes, "No, it's a skill issue." Apparently that's the lingo right now. He was saying it's a skill issue if you can't get value out of these models. There's this idea of token maxing, which is a very technical concept, but it actually makes a ton of sense. Every time you use one of these models, you're basically burning through tokens. Tokens are like pieces of words in essence. You get an allotment of these tokens. Let's say I use a million, 2 million tokens, whatever. He was saying if you're an engineer, you want to know what your token budget is. How much AI can I use in my job?

This idea of token maxing is like for the average user like you and me, Mike. I have a Claude license, I have a ChatGPT license, and I have a Gemini license. If I'm not maxing out my subscription every month, I'm leaving intelligence and outcomes on the table. He was saying there's this pressure right now, especially on coders, to max out your available tokens. Because if you don't, you're just not getting the full value. I think that concept is going to start carrying over eventually into knowledge work where you're like, we have these AI tools, we're not fully utilizing them, and we're just leaving value on the table by not maxing out our tokens each month.

In a similar place, he talked about this idea of running projects in parallel, which I do. I'll go into Claude and say, "Okay, I'm going to give it this project." And I'm going to go over to ChatGPT and I'll have it work on this project. There are times where I'm running three projects simultaneously with AI agents while I'm doing my other work. I'm doing email or I'm doing something else and I've got these running. That's a big thing. Then he talked about the compression of timelines to complete projects, which I'm going to talk about in an upcoming topic here about our company retreat. I think that's a very important concept—that things that used to take 5 hours, 10 hours, 20 hours now might take 5 minutes. That's a weird environment to be working within.

He also talked simultaneously about this idea of compression of software stacks, where we used to have a CRM tool and a social tool and all these tools. And it's like, "I'm just going to have a swarm of agents and they're going to go talk to all this software. And I'm just going to have a single user interface." The final one I'll say with Karpathy—and again, this is all relevant to what these labs are doing—if you listen to the Karpathy interview, all the labs are realizing what Karpathy is realizing on agentic capabilities. They are now in a race to do what he explains in this. That's why this podcast episode is so important, that No Priors episode. He is telling you point-blank what all the labs are trying to do with agents. You will walk away with a better understanding of the moment. He said at one point, "Working with these agents is like simultaneously talking to a PhD student and a 10-year-old." Sometimes you do something with it and it's like giving it to a top PhD student. And then the next moment, it's some stupid simple thing and it just can't do it. It's that idea of the jagged frontier and the jaggedness of these models.

Zoom out, recap. All the labs realize what Claude Code unlocked. And it wasn't like it was the first coding agent. It was just the best. They did something different with the harness, like how they enabled it to do what it does. All these labs see not the finish line, but the next mile marker of agentic capability and their ability to automate AI research and their ability to then, as Logan Kilpatrick's deleted tweet said, to start disrupting everything. It is an all-out race for agents. They're seeing a pot of gold with enterprise adoption, which is why Anthropic and OpenAI are doing deals with PE firms. It's why they're doing alliances with major consulting firms. They're trying to get in and get where this is going to be because the labor replacement value of being the model they go to when they reduce workforces and put it all into AI models to token max to get work done—they see that future coming very fast. It's important that you understand what we just covered. That's what these labs are doing. It's going to become very apparent in the next 3 to 6 months that this is full go where they're headed.

[Mike Kaput]: Probably a pretty good time to be an enterprise buying AI technology. I'm assuming these labs would like to court you.

[Paul Roetzer]: Yeah, you get a lot of credits, especially here. Like, I'll give you the first million free.

New Polling on AI and Trump National AI Framework

[Mike Kaput]: All right. Next up, we've got three separate developments this week that are painting an increasingly complicated picture of how Americans actually feel about AI and how Washington is responding. First, we had some new polling. David Shore, who is head of data science at Blue Rose Research, appeared on the Odd Lots podcast with some interesting polling data. His organization has found that over the past year, AI rose in issue importance faster than any issue his firm tracked. It is now more important to voters than climate change, child care, and abortion. According to their polling, 79% of voters are concerned the government doesn't have a plan to protect workers from AI job losses. 77% are concerned about entire industries being eliminated. 56% are worried about personally losing their job to AI.

This is hitting at a time when 61% of Americans say life has gotten less affordable in the last year. Only 25% feel confident in their financial future. Only 34% say that, in their opinion, they have a secure job. What Shore's data shows—and he's polling from this perspective of trying to find political messaging for the Democratic Party—is that this whole idea of, "Hey, everything's going to work out just fine," that message is dead on arrival. They actually found when leaders in government and tech say AI will not cause widespread job losses, net trust is negative 41. And when they say AI will create economic productivity that benefits everyone, net trust is negative 20.

You're starting to see this play out across the political spectrum because we've gotten dueling AI political declarations. First, there was a coalition that involves a lot of unlikely bedfellows, including Steve Bannon, Susan Rice, Richard Branson, Ralph Nader, Yoshua Bengio, and others who released the pro-human AI declaration. This basically called for a prohibition on superintelligence development until there's broad scientific consensus it can be done safely, as well as a number of other manifesto points about keeping AI pro-human. Over 40 organizations signed this, and they also found in their own polling that Americans would rather prefer human control over the speed of AI development by an 8:1 ratio.

However, another organization called Build American AI published a direct counter to this manifesto titled, "We Cannot Afford to Pause AI." They argued safety and innovation are not opposites, and the US already has regulatory tools through existing authorities to manage AI development.

Third, the Trump administration unveiled a national AI legislative framework with seven pillars. This is a short document, but basically gives legislative guidance on how they think legislation should evolve related to AI. This framework takes a pretty clear try-first rather than regulate-first posture. It opposes creating any new federal AI regulatory bodies. It defers copyright questions to the courts rather than legislating. And it recommends Congress preempt state AI regulations that impose undue burdens on developers, establishing what it calls Americans' right to compute. There's an interesting part in here in shifting responsibility for protecting children online from tech companies to parents. Rather than imposing strict industry standards, they are actually shifting more to empowering parents with tools to protect kids online. The framework also calls for Congress to empower Americans to challenge federal agency efforts to dictate the information provided by an AI platform. Basically, they are trying to make sure that there is no undue influence on what information is provided by AI. Paul, I'm curious, there's a number of threads going on here. If you're in the AI industry or just observing or trying to navigate these changes yourself, how are you thinking about these numbers and the moves on either end of the political spectrum?

[Paul Roetzer]: Like any research, we always talk about how you have to know who's doing the research and what their goal is, and what kind of bias might be in the research. That being said, it's going to become more political. As we've said many times in recent months, this is all trial balloons. They're trying to figure out what Americans think about AI, and is there an opportunity to move votes a few percentage points one way or the other by taking a strong position on AI, which Republicans and Democrats haven't really for the most part with voters.

The one thing that is becoming more interesting to me is—I always read this research and think these people don't know what AI is. You're asking them questions about something that they don't understand. And now I'm actually thinking out loud here—that is maybe an advantage for politicians that want to manipulate and persuade people to vote one way or the other. If you don't know what it is, then you can create the perception of what you want. If people generally are like, "I don't know, whatever," then it's like, "Okay, let's hammer the message of it's going to take jobs, and data centers are going to ruin communities." And now that's all AI is to people. This is a dangerous slope we're going down here where we're seeing the early efforts to try and gauge what is perception so that we can then influence perception of what it is to move votes one way or the other.

David Shore—I didn't know who he was. I didn't know his organization. So, that's always the first thing I do. We see some cool data, and it's getting shared everywhere on X. Who are these people? What is their mission? David Shore is head of data science at Blue Rose Research based in New York, originally from Miami. "I try to elect Democrats." That is his X profile. There's no hiding what the point of this is. Blue Rose Research helps campaigns make higher quality strategic decisions by democratizing access to accurate measurement. That's on their about us page. The name Blue Rose symbolizes turning blue what is now red. Again, there's no hiding what this is for. David Shore is a prominent American data scientist, political consultant, and expert in public opinion polling. Now, that doesn't mean it's not valid research. We're just saying there's a perspective here. That's the whole point of understanding this. He actually worked for Barack Obama's 2012 re-election campaign.

The survey, just to put it in a little bit of context—when it says AI is the fastest-growing issue, you have to understand it's actually 29th out of 39 issues right now. Yes, it's growing fast, but the top five issues for Americans are cost of living, the economy, political corruption, inflation, and healthcare. Those don't really move. Those are pretty common top five. Then if you go down to 25 to 30—just to put in context of where AI falls—you have war in the Middle East at 25, international trade, income inequality, voting rights, then artificial intelligence, then race relations. While it is growing fast, on the surface, Americans don't really care. It is not something that would jump out to you as like votes are going to move based on that.

But it is changing fast. You talked about some of these key ones, Mike. The government not having a plan to protect workers from job loss. The question was, "How concerned are you about the government not having a plan to protect workers from job losses driven by AI?" 79%. You don't need to understand what AI is to be like, "Yeah, it kind of worries me that there isn't a plan." And that is 100% true. They do not have a plan. Or if they have a plan, they're certainly not talking about the plan. Everyone should be concerned that the government doesn't have a plan. Then it said, "How concerned are you about young people entering the workforce and finding fewer job opportunities because of AI?" 79%. They should be concerned. That's happening. That is a real thing right now. Whoever's asking these questions—Republican, Democrat, independent—doesn't matter. That is a fact. It's harder to find jobs right now.

"Entire industries being eliminated by AI faster than new ones are created." That's a ridiculous question. We're not getting rid of industries. Companies being disrupted, sure. Career paths—that's an absurd question. You could just throw that one away. "AI changing the job market in a way that drives down wages for people like you," 72%. You could replace AI with any variable. Anything you ask is like, "Are you concerned with something driving wages down?" Well, of course I'm concerned. I don't want my wages going down. So, it's like, whatever. "You or someone in your family losing their job in the next year because of AI," 56%. That's a reasonable concern. When they say when leaders in government and tech industry say AI will not cause widespread job losses, net trust, as you mentioned, is negative 41. Distrust it somewhat, 35%. Distrust it completely, 32%. So, 67% distrust it somewhat or completely. Now, that may align with 67% of people don't believe anything government tells you. I have no idea.

There was another one, Data for Progress, which is a progressive think tank and polling firm that provides data, research, and messaging strategies for the progressive movement. They produce polling on policy issues and support campaigns. They came out with new research on February 27th, which is worth mentioning here. This is 1,200 US likely voters nationally using a web panel. They're asking about how frequently they use AI in their daily lives, whether they have favorable or unfavorable views of the tech, and how confident they are in their ability to spot AI-generated content. It's a pretty short survey. We'll put the link in. It's only like five pages. You can read it for yourself if you want. But some of these questions are pretty interesting. "Do you have a favorable or unfavorable opinion of the following people or institutions?" They asked about AI. Democrats minus three, net favorable. Republicans plus 11, independents minus five.

They asked, "When it comes to AI tools such as ChatGPT?" Now again, they're trying to qualify for you what we are talking about when we talk about AI. If you understand what ChatGPT is, at least you have some concept. "When it comes to it in your personal life, have you mostly embraced or resisted using them to assist your life? Or have you found areas where you could use AI in your personal life?" Embraced: Democrats 32, Republicans or the press 34, Republicans 32. Resisted: Democrats 35, Republicans 33. "I have not found areas that I could use AI in my life": Democrats 30, Republicans 32. It is totally balanced. There's really nothing there that would indicate any anything they can do with that data or to move people one way or the other.

Then they had another one: "Sometimes people use AI to make fake or edited photos and videos that they post online. How confident do you feel in your ability to spot that stuff?" Very confident 15%, somewhat confident 35%. So, 50% think they can figure it out. They're wrong.

Then they did an interesting one where they were comparing data from August 2025 to February 2026, where they asked how frequently, if at all, do you use AI such as ChatGPT for your job? Right now 14% say multiple times a day. 44% rarely or never. 11% a few times a month. You have 55% of these people being polled in February 2026 that use it a few times a month, rarely, or never. If you think everybody's doing this, they're not.

The one you mentioned about the pro-human AI declaration—again, it's important to know where the counter's coming from. The AI industry super PACs—we talked about this last year. CNBC had this as well as others. There's a super PAC called Leading the Future. The contributors to this are Andreessen Horowitz, OpenAI co-founder Greg Brockman, Palantir co-founder Joe Lonsdale, SV Angel founder Ron Conway, and AI software company Perplexity. These are people pushing this super PAC, which is all about acceleration. It's all about rapidly accelerating what's going on. They're basically saying that this stuff is ridiculous. The Build America AI is in essence led by this group. They're saying we cannot afford to pause AI. This TechCrunch piece highlights the release of the pro-human AI declaration, the document you mentioned. The goals behind that effort are understandable. People want AI to be safe, and they want clear rules. Those are fair concerns, but this is still the wrong direction. This is the super PAC people. Pausing frontier AI development will not solve the problems its supporters claim it will solve. If anything, it risks making several of them worse. It would slow the research that helps us understand how these systems behave in practice, and weaken America's position at the exact moment our adversaries are investing heavily in advanced technology. We cannot hand hostile actors on the world stage a strategic edge. That is what would occur if we paused AI.

That leads to the AI legislative framework from the government, which is just the starting point. That's the most important thing to take away from that. It's just like guidance on where they think legislation should go. It's not doing anything. But you covered some of it. It's like protecting children, safeguarding and strengthening American communities, respecting intellectual property rights. That's a really funny choice of words—respecting intellectual property rights, meaning they don't want you to have property rights as a creator, and supporting creators, preventing censorship and protecting free speech, enabling innovation, and ensuring American AI dominance. That's probably the most important one, because all the other ones fall under that one. And then educating Americans and developing an AI-ready workforce, which I'm definitely intrigued to hear what they've got in mind there.

Again, it's just—I think what we're seeing, we've said it recently, every week there's now going to be more and more on the political side. We are moving into the midterms. We are moving into the moment where the political parties have to decide whether or not Americans care. This election cycle is either going to be all about AI, or it's just going to fade away. You're seeing the push towards data centers being bad, job loss being bad. And then you've got the Leading the Future super PAC people who are like all of it's great, and it's all going to create an abundant future for all of us. And if you don't believe that, then believe we have to beat China. That's basically the messaging. It's like choose your fighter. I don't know where the middle ground is here, but right now neither side really knows. But the super PAC, the Leading the Future people, are going to push hard on this stuff, and they're going to try and make you believe it's all going to work out, and jobs aren't going to be lost.

What I would just encourage people to do is don't get stuck in whatever your traditional political silos are. If you only listen to one perspective on this—this is an issue where you can't just be listening to one perspective that you've always followed. I think it's really important to realize neither political party knows the answer here. They're both trying to figure it out. It's really important that you open your own mind and look behind who's saying things and what the goal they have behind saying that is, or where the research is coming from. It's going to be very important to try and keep a level head on this stuff and listen to arguments of both sides.

[Mike Kaput]: To your point about people often being polled who don't know what AI actually is, that's the point of some of these numbers. We would throw out half these questions if we were doing actual research. But if they surface a strong opinion or view on AI, even if that view is wrong, that's really useful polling to certain people, because it tells you exactly what you need to say and hit on, using that ignorance almost as a weapon in some ways.

[Paul Roetzer]: Yes, facts and lies mean nothing in election cycles. It's all about what you can say that'll get you to remain in power. That's not a controversial statement. It is what it is. They can tell you whatever you want to hear to stay in power or to get in power. Both sides. Form your own opinion. Form your own informed understanding of the situation. And then from there you can take more logical actions to make sure you understand. It's like situational awareness about what's happening with this issue. It's going to become a major issue, I think. I think they're going to find the levers to pull. They're going to find the wedges to create frustration and anxiety around AI, and that could get very dicey.

Company Transformation with AI (Offsite Recap)

[Mike Kaput]: Our third big topic this week is about the SmarterX annual meeting and retreat we had over the last couple days of last week with our team. This was super inspiring. We spent a couple days together collaborating. Day one, we talked about vision, goals, KPIs, priorities, and growth initiatives. Day two, we ran AI productivity and AI innovation workshops, which are designed to accelerate responsible AI adoption across business units and teams. The reason we wanted to cover this and dive into it is because it has some signals, maybe some lessons here about overall company transformation with AI. Paul, I'll let you unpack this for us, because what we were able to achieve over just two days, both in how we were approaching AI and by actually using the technology, I think can teach us quite a bit about how AI is changing the way businesses operate.

[Paul Roetzer]: A couple things. Mike, you and I haven't talked about this. So, if you have other perspectives or things to add, let me know. But the reason we wanted to highlight this is a few things came to me. It was two days. There was a part of me that thought it was a great example of what you can do with the time you gain from AI. The fact that we use AI so intelligently within our own business gives us a little freedom to say, yeah, let's take a full two days. Let's go do this thing. Let's go think. Let's go spend time together, build camaraderie. Do all the things we should be doing. As I was sitting there, I kept thinking we have to do more of this. When I think what an AI-forward company looks like, and how you take the benefits you gain from AI—the efficiency and productivity gains—and redistribute that in some way? I'm not a four-day work week guy. I don't think that's reality. I do love the idea though of, let's do more of this. Let's have once a month where we just take an afternoon to just think and talk and work on big ideas. I find that enables the work to be more fun and more fulfilling if it's not just let's token max every minute of every day. In some ways I want to build—I want to maximize what we can do, but I also want to make sure we're getting the benefits of it. It's not like a race to some end game or some competitive race.

The way we set it up was day one was the company day. It was vision, goals, KPIs, building scorecards, a Rocks workshop where we're setting priorities for the coming quarter. And then I think just like the thing we teach, which is setting expectations for everyone of what an AI-forward professional looks like. And in some ways modeling that by showing in real time how we're using AI, and making sure everyone on the team understands the capabilities. Mike, you did on day two, you led off with this AI productivity workshop. And you talked about the idea of not only jobs as tasks, but tasks as workflows, which I loved that framing. And then you went through an AI capabilities overview of what all these things the models can do, so that people started to think a little differently about their own daily lives at work. We demoed Jobs GPT, Campaigns GPT, and Innovations GPT. I did that one in mine. But those are some of the free custom GPTs we've built that we make publicly available. We use them in our own teams. We literally use these tools to train our own teams as an example of this AI-forward idea in real time.

Mike's doing his workshop, which was awesome, because I've never sat through one of Mike's workshops. Mike and I do these things all the time for other companies. We do them at our MAICON event. But we don't have time to sit in each other's workshops. He's doing this workshop, and he's showing AI layering over workflows and reimagining workflows. He showed this AI capability slide, and then he turned it into it was like a spreadsheet with 90 rows or something, right? Like 90 different capabilities and features across some of the major AI tools. So you can quickly pick and choose and filter and map things to all the individual tasks you're doing as part of a workflow, for instance. Reasoning capabilities, video capabilities.

I loved this. And I was looking at this thing, and I wondered if we could turn this into something. As he's talking, I take the spreadsheet and I put it into Claude Code—or just Claude, and I'm using Sonnet 4.6 at this point. I said, "Help me visualize this. We want it to help professionals understand the full capabilities of today's leading AI models so they can apply them to their work." That was the entire prompt. It did it. I thought this was really cool. I said, "Is there a way to turn this into an app that I can demo internally?" In about 3 minutes later, I had this functioning app. Mike doesn't know this is happening. He's just on stage doing his thing.

But the best part—and this I'm still trying to wrap my head around this—Mike, we don't have a Claude license for the team. So when Mike built his capability slide, it was what? Google?

[Mike Kaput]: I said Google Gemini, ChatGPT, Notebook LM, and I spun out kind of deep research for both tools as kind of its own capability set.

[Paul Roetzer]: Okay, so this 90-row worksheet does not have Claude in it. But I'm talking to Claude to build this interactive demo. Claude says—it first asked me a question, "How would you want to share or run it?" And I said, "Standalone HTML file's fine." It then said, "What should people be able to do beyond browsing? Select all that apply." And I just simply said, "Just browse. That's enough." Then this is the question that blew my mind. It said, "Should Claude be included as a fourth tool?" It was aware that it wasn't part of the spreadsheet he created. And it asked me if it should add itself to the spreadsheet. I literally laughed out loud when I saw this. I said, "Yes, add Claude." And it did. And it followed the exact model he had done for them. And then it built this interactive capability thing. It honestly blew my mind. We do this stuff every day. I see this stuff every day. And there' still moments where I can't even believe it was capable of doing this in real time. When I say Claude is running circles around what some of these other apps are capable of doing, this is a perfect example of it.

[Mike Kaput]: You one-shotted a 90-item capability database—more than 90, because it added in probably 25 different things from Claude. One-shotted it in a way that genuinely was professionally designed, extremely intuitive. It was great. It was search functions, filter capabilities. Unbelievable.

[Paul Roetzer]: The other one I'll share—and again, we'll touch on some of this later on—was rocks. I put this on LinkedIn on Sunday, and I actually featured this in my newsletter, the Executive AI Insider newsletter. I'm just going to read what I wrote because it summarized it really well. I was saying we went through this retreat, and one of the things that became apparent to me as an example was this idea of rocks. We use a modified version of rocks from the EOS system, in which departments and individuals establish three to five priorities per quarter, and then the rocks allow us to align our time, energy, resources on what matter most, and it provides transparency. If I want to see what are the five things Mike's working on in Q2, I can go look. Or if I want to see what the studio that Mike leads is doing, I can go see that.

The thing that became abundantly clear to me is the time to complete rocks is compressing, and that it requires a complete rethinking of business operating systems. For example, during the live session where I was actually demoing—this is part of the company day—I was demoing a new AI assessment tool we're developing that I'll share more about in probably a month or two. And I'd used Anthropic Claude Code, again Sonnet 4.6, in real time to build an interactive reporting dashboard that visualized and analyzed responses from 17 people. I had built this assessment in Google Forms as an MVP, and then Mike and I tested it the day before the retreat just to make sure it worked. I had my data and Mike's data, and then I had everybody else take it, and then I exported that CSV from Google Sheets. That was it. That was the entire process. Zero coding, zero design ability to do this thing. I give this to Claude, and I said—this is while we were taking a lunch break, I ran this. Here's my prompt: "I had 17 team members take the assessment. Can you come up with an elegant way to visualize the results based on the format model you already created?" I had it create one for me and Mike. "And the CSV is attached."

In a previous life—aka 3 months ago before Claude Code really started working—this would have been my entire Q2 rock. Create an interactive dashboard to visualize assessment results for teams. I would have spent 10 to 20 hours researching dashboards and developing a brief. Then I would have invested time and money hiring a designer and developer to conceptualize, build, and iterate on the design and capabilities. Then we would have gone through weeks of internal testing and revisions. And then maybe by the end of Q2, I would have actually had a minimum viable product that I could demonstrate to the team and pilot with users. Instead, in about 5 minutes while I got a plate of pasta, Claude did the entire thing with one prompt, and the final product was beyond anything we could have possibly created. I told Mike, "I'm going to try this. I'm going to do it." And then he and I are both just waiting. We go check the laptop. Did it do it? It was insane. It was totally interactive, better than anything I could have possibly designed myself or worked with a developer to build. I'm now going to use that to actually turn it over to a developer and say, "Here, let's build this and take this live in 30 days, hopefully."

We shared this as a little bit behind the scenes of how we think about SmarterX as an AI-native company—event and media and education company. And two, just to bring to life the fact that you don't need any coding ability to all of a sudden now just build stuff. It's totally compressing the timelines to do everything in business. It's changing the way every day that I think about how to run our own company and how to advise other people to build their companies.

[Mike Kaput]: I would argue we have quite well-done and clear and ambitious rocks, at least in our department that we were working on during this workshop. But yeah, it is actually kind of laughable that all five of them should take 3 months.

[Paul Roetzer]: My guidance to the team was like five. I want you to have five for your department. I actually think you need 20. And you need some categorical thing of like, "Hey, this is something that would take 10 to 20 hours of human labor. We think we can do it in 10 minutes." There are honestly things that are just going to be like that. There's going to be all these quick win rocks where it's like, "Well, it used to be 3 months worth of work, but it's probably 3 days now with mostly AI. It's like level three AI. It's going to do most of the work." The companies that figure that out and realize that and restructure how they're building everything stand to do really well.

[Mike Kaput]: Just two final really quick notes here, but to piggyback on what you did with AI during these workshops. The AI capabilities map I built, which was 90-ish rows of all these different capabilities and features—that's a lot to figure out on your own. What's really cool is I determined the framework I wanted to use and worked back and forth with Claude to say, "Okay, what's the most sensible way to organize these once we have them?" I don't have them yet. And then it's like, "Okay, we've got a really solid system. How do I get them?" Typically, you might go do a bunch of research. You might have to sort through all sorts of documentation. I just went into each tool and screenshotted all my menu options. And then dropped them into Claude and said, "Guess what? We're going to go create the spreadsheet based on the framework that you and I came up with, and go have at it." And then it basically one-shotted a 90-row spreadsheet. It's incredible.

And same type of thing during your innovation workshop. I fed Claude a lot of different context about my department, Content Studio, some of my context around our organization, and then using your framework that you developed, layered that on top of that context and what Claude is now able to do, and got better innovation ideas than I could have come up with first on my own at all. And second, in an entire day, I did it in like 20 minutes. It's so powerful, not only just using the right tools, of course, but having these proven frameworks and models and ways of thinking layered over them. All that stuff we've spent lots of time developing as IP or as unique models to approach these things with in our workshops—it's like rocket fuel at this state.

[Paul Roetzer]: I think there's just something like a lesson to be taken from how we structured it because obviously our team's probably more informed than most teams about AI capabilities, but honestly, I don't know that they even were aware of a lot of the things these models could do. It was very intentional how we did this, and I would advise other companies to think about a similar model where you have this kind of state of AI—what is it capable of? That's often what I'll go in and do with enterprises. I'll do a state of AI for business, and here's the capabilities, here's what you need to understand. Then you do the productivity workshop where it's like, "How do we get efficiency and productivity in our tasks and workflows?" Then we'll often do a problem-solving one, too. But the innovation one is how we close, and I intentionally wanted to close with that because once you understand what it's capable of, and once you've solved the lower-level efficiency and productivity things, now you open your mind to the possibilities. And then we go around the room, and each person gives us one or two innovations they're super excited about. So then you leave after 2 days actually feeling ready to go, not drained. It's like, "Okay, that was amazing. I want to go do those things now." People were coming up to me, like, "Okay, can we do these things that we just talked about?"

I think it's a really cool format for people. If you're trying to get your team on board, borrow that format of making sure they're understanding of it. And if you need help with it, give us a call. This is what Mike and I do all the time. We run boot camps and workshops. If nothing else, we can advise you on ways to do it. But if you're a big enterprise and you need help with it, we can come in and do stuff like that, too.

Nadella Takes Over Microsoft Copilot

[Mike Kaput]: All right, Paul. Before we dive into rapid fire, quick message here. This episode is also brought to us by our upcoming webinar, which is unveiling our AI for CMOs blueprint presented by Google Cloud. This is actually happening the week you are listening to this episode, Thursday, March 26th at 12:00 p.m. Eastern, 9:00 a.m. Pacific. In this session, me and our CMO, Cathy McPhillips, are going to break down the insights from this AI for CMOs blueprint we put together in partnership with Google, where we break down real world state of AI for CMOs, use cases, tools, strategies, and more. We'll also be doing some in-depth discussion and live Q&A. Registration is free. All registrants will receive ungated access to the full AI for CMOs blueprint. So, go to smarterx.ai/webinars to go register.

All right, let's dive into rapid fire, Paul. First up, Microsoft CEO Satya Nadella is taking some more direct control of the company's Copilot product, personally overseeing a restructuring that consolidates consumer and commercial Copilot into a single organization. Jacob Andreou, a former Snap SVP who joined Microsoft last year, now reports directly to Nadella as the new EVP leading Copilot experience across both segments. The restructuring frees up Mustafa Suleyman, the DeepMind co-founder who became CEO of Microsoft AI in 2024, to focus entirely on what he calls the company's super intelligence efforts.

This move apparently comes as Copilot trails quite badly in the AI assistant race. Copilot has 6 million daily active users compared to ChatGPT's 440 million, that is according to a CNBC article. Gemini has 82 million, Claude has 9 million. Nadella wrote to employees that Microsoft is doubling down on our super intelligence mission with the talent and compute to build models that have real product impact. Paul, what does this tell you where Microsoft is headed with AI? Reading this, I thought it sounds like Mustafa is excited, but this feels more like he's getting sidelined and we need to get real serious about Copilot real quick, which is kind of what we've heard anecdotally from users of Copilot.

[Paul Roetzer]: There's lots of variables going on here. One is the shift in their relationship with OpenAI. They were obviously a major investor in OpenAI. They're a major equity holder. I think it's somewhere around 27% they own of OpenAI. But all of their efforts were being built on top of OpenAI's models. And now, again, if you go look at what we were just talking about with Claude, it's like you're almost at a disadvantage as an organization if you can't use when a breakthrough happens, when somebody builds just a better thing, you're at a disadvantage if you can't use that thing. If Microsoft was stuck using OpenAI technology and all of a sudden Claude races ahead in some really important component, that's not great. And then if you're Microsoft and you're one of the three biggest companies in the world, the fact that you aren't building your own models is probably a disadvantage moving forward. I think there was the shift where they realized that probably a year and a half, two years ago that they were going to have to remove their reliance on OpenAI. It probably happened the day Sam got fired when that became like, oh boy, we are all eggs in one basket and it could go bad real fast. I think there's been this ongoing shift where they knew they needed to invest in their own technology, build their own models. They need to have an off-ramp over time from their reliance on OpenAI.

In November of last year, they announced this humanist super intelligence movement. We talked about it in episode 179, which was on November 11th. Mustafa had tweeted, "It shouldn't be controversial to say AI should always remain in human control, that we humans should remain at the top of the food chain. That means we need to start getting serious about guardrails now before super intelligence is too advanced for us to impose them." And then there was linking to an article from November 6th that was called Towards Humanist Super Intelligence, where he said, "At Microsoft AI, we're working towards humanist super intelligence, incredibly advanced AI capabilities that always work for and in service of people and humanity more generally." We've kind of known this was happening. At that time, I think I pulled what I said. I said, "Maybe Mustafa stays at Microsoft to realize this vision, but I can't help but feel like this vision will eventually clash with the need to justify their investments in AI."

I think what they're basically saying is you go focus on this stuff, focus on the future and the building of this thing, but Copilot is critical to our business right now and it is not where we want it to be. And that now needs to get much closer to Satya. That's basically what has happened here. I have no idea if Mustafa stays and keeps doing what he's doing, if they really do believe in this humanist super intelligence thing, but I don't see Wall Street loving the humanist super intelligence. I don't think the stock price is going up because of that blog post or that vision. They want to know how you're going to compete with Claude and work with Anthropic. And that's all that Wall Street's going to care about. At the end of the day, Satya and Microsoft have a fiduciary responsibility to return shareholder value. I don't think that messaging plays. We'll see. It fits into that whole thing I started off with where these AI labs are shifting focus. You're going to see a lot of reorgs, a lot of like, they tried something and it didn't work. Meta burned 10 billion on the metaverse and changed their name to be Meta, and it's done. There's going to be lots of big efforts, big misses, and you have to move quick when it doesn't work. I think this is an example of that.

[Mike Kaput]: Not to mention anyone who's a Wall Street analyst of any type is almost certainly using Microsoft Excel and thus Copilot and sees it firsthand. Or they used Claude in Excel and realized it was better than Microsoft Copilot. They have very close experience with perhaps some of the inadequacies.

Meta's Rogue AI Agent

[Mike Kaput]: All right, next up, an AI agent inside Meta took unauthorized action last week that triggered an actual security breach at the company. An employee used an in-house agentic AI to analyze a colleague's question on an internal forum. Pointed the AI at the question, said, "Analyze this for me." The agent then posted a response to the question on its own without being directed to do so. The second employee followed the agent's advice, sparking a domino effect that gave some engineers access to Meta systems they should not have been able to see. The security breach was active for 2 hours before it was contained. A Meta representative confirmed the incident and said no user data was mishandled, though the company's internal report noted unspecified additional issues that contributed to the breach. A source told The Information there was no evidence anyone exploited the unauthorized access or that data was made public, though the reporting notes that may have been the result of dumb luck more than anything else. The agent had also passed every identity check in Meta's system. That exposes some pretty serious fundamental gaps in enterprise identity and access management. Paul, I'm curious, how close are most companies to having this kind of thing happen to them?

[Paul Roetzer]: I don't know, but it's certainly a very viable thing. This is why I said in a recent episode, you have to listen to IT. There's a reason why some enterprises are moving really slow, especially when it comes to adoption of agents. When Jensen was like, "Open Claude is like the ChatGPT model," I thought, "Okay, maybe, but you know how hard it's going to be in enterprises to do anything close to what that does." This is the exact issue. We just talked on episode 203 about something similar happening with Amazon, where it just went rogue and started doing everything. I think I joked at the time like we could just do a rogue AI agent segment every week. This is going to be a recurring theme. It's going to become a major issue. The concerns around oversight and governance of these agents and then these agent swarms that are just given access to stuff and the breakdown you might then see in permissions controls.

We had this conversation at our own company meeting. It's like, can we connect this to that? Can we connect that to this? And it's like, no, because I don't know yet the risks associated with that. This is a crazy one. You should go read the articles about it. It's pretty nuts.

[Mike Kaput]: I almost found this was more notable because it happened—it wasn't like some super incredible agent just giving access to your whole code base or whatever. It was just a totally unintended consequence of something that's actually a pretty normal use case on the surface saying, "Hey, let me use AI to analyze a question one of my colleagues posted on a forum." And then you're like, "Oh, no. I realize that now this thing can choose what to do and how to do it." And that's a totally weird way to start thinking here.

[Paul Roetzer]: Yeah, and again, go listen to Andrej Karpathy's No Priors podcast episode and you'll understand this stuff at a deeper level. He talks a lot about these risks and even himself not knowing. He talked about setting it up to run his house. He was like, "Oh yeah, I gave it access to—I was like, go find Sonos," and it goes into his network and finds the Sonos speakers and then he goes to the security cam and he just gave it access to everything and he calls it Dobby the home app. It's hilarious. So again, this is a recurring theme. It is really important to understand where agents are going, where these agent swarms are going, how they'll eventually be used to run organizations, and how some people are willing to be out on the edges right now setting these things up and connecting them to their own company data. We're all going to learn plenty of lessons from their early efforts.

Anthropic vs. Pentagon Continues

[Mike Kaput]: All right, in our next rapid fire topic this week the Anthropic vs. Pentagon saga continues. The Department of War has fired back at Anthropic's lawsuits in a 40-page filing in California federal court. The Pentagon calls Anthropic an "unacceptable risk to national security," arguing the company might attempt to disable its technology or preemptively alter the behavior of its model during war-fighting operations if its corporate red lines are being crossed. Recently, nearly 150 retired federal and state judges appointed by both Republicans and Democrats have also filed their own amicus brief supporting Anthropic. We talked last week about how tech companies like Microsoft and Apple have all filed their own briefs basically arguing that this designation of Anthropic as a supply chain risk could mean the entire government procurement system becomes contingent on political favor rather than the rule of law. Paul, the big piece here is really this idea that a bunch of ex-judges are coming out and saying that they also support Anthropic in this. We have talked about if this is going to get resolved anytime soon. There's a hearing on whether or not to grant Anthropic some temporary relief that's actually set for March 24th, the date this comes out. Where do we stand with this?

[Paul Roetzer]: I don't know all the context. I think it's just still this he-said, she-said thing where the government's saying one thing and they're doing the other thing behind the scenes, but they're trying to give this perception that they're all in the right here and Anthropic is this horrible company and it's this huge risk. There was a tweet thread from Roger Parloff, who's a senior editor at Lawfare, and I'll put the link in. He said some Anthropic updates on March 4th—just hours before Hexeth declared Anthropic a supply chain risk allegedly due to threats of sabotage and data exfiltration, his undersecretary wrote Anthropic, and he had the screenshot of the email, that they were very close to a deal asking to change a prepositional phrase. So while Hexeth's getting ready to go on and blast them on X and say they're done, they're actually still negotiating behind the scenes and they have screenshots of it.

Since then, the government has claimed that Anthropic sought a veto over Department of Defense actions, but two top Anthropic officials assert it never did, and this is actually legally—they submitted a briefing saying this is not what happened. Similarly, the government's purported fear that Anthropic might disrupt the military was never raised with the company and is a technical impossibility, so they actually explained like we can't even do the thing they're claiming we would do. And then as for Anthropic's refusal to allow its product to be used for autonomous lethal warfare and mass surveillance, Hexeth himself said those concerns were understandable and the commander of the US CENTCOM echoed those sentiments to Anthropic's head of policy. They submitted these briefs saying they agreed with us—we weren't even raising something that they didn't themselves think was an issue.

In the government's response Tuesday, it backed away from the secondary boycott Hexeth called for in his February 27th "final decision" post on X, admitting it was lawless but also taking no responsibility for its devastating impact. The hearing is coming up on March 24th. These are declarations—legal declarations from Anthropic's head of policy Sarah Heck submitted as part of their response to the case. And then they also had their head of public sector. They're basically saying, "Well, here, I'll testify to like this never happened or this is what they said." The whole thing, as I've said many many times, has become this political thing. It's become a battle of egos on the government side and I think that everyone sort of sees through why they're actually doing this. We'll see what the courts have to say, I guess.

[Mike Kaput]: It's impossible to tell, but based on that new context, it almost sounds like there's one possibility where Hexeth jumped the gun on tweeting about this when they were nearing a deal.

[Paul Roetzer]: Well, I didn't say he jumped the gun, but claimed some power that they actually don't have, where he's posting so aggressively when the deal's almost done before this all blows up. Now it's just doubling down on a mistake, maybe. Or just you're just going to do harm either way so you don't really care if it's legal or not. He doesn't consider repercussions—nothing's going to happen to me if I do this and say this other than hurt that company and try and use it as leverage to get them to do what I want them to do, which is not an unusual political tactic.

DeepMind’s New AGI Scorecard

[Mike Kaput]: All right, next up Google DeepMind has published a cognitive framework this week that attempts to answer the question: if AI actually achieved AGI, how would anyone know? The team here proposes a cognitive taxonomy with what they claim are 10 measurable traits of general intelligence on which to measure AI and its progress towards AGI, and it's divided into two categories. This first category covers eight building blocks of human cognition: perception, generation, attention, learning, memory, metacognition, and executive functions. These combine to form two composite faculties that DeepMind considers equally important, which are problem-solving and social cognitions. They basically define these as the ability to process and interpret social information and respond appropriately in social situations. Their proposed test here is pretty straightforward. They want to run AI models and humans through the same cognitive benchmarks and then they theorize you'd get a measurable estimate of when a single AI can meet or exceed human capabilities across all 10 of these areas. DeepMind actually launched a Kaggle hackathon with a $200,000 prize pool to crowdsource evaluations for the five areas where the gap between testing capabilities right now is the largest, which are learning, metacognition, attention, executive functions, and social cognition. They say their goal is to move the conversation around AGI from one of subjective claims and speculation towards a grounded, measurable scientific endeavor. Paul, does this change anything about how we talk about AGI? Are we getting any closer to really defining what it is and actually measuring it?

[Paul Roetzer]: Google's DeepMind's done the best job of trying to get to that point. They had a paper last year where they were trying to sort of define the different general capabilities and performance and trying to put some way to measure it. I like the effort to try and quantify it, making it more meaningful, trying to get some eventually universal agreement on what it is. The first thing I thought when I saw this is—how do you not saturate these tests? When the models eventually learn what the tests are and just be able—I don't know how they would do that to keep them sandboxed so the model doesn't end up in the training data, basically, so that it eventually learns how to look like it has AGI because it just learned what the test was ahead of time.

But I think the most important thing for our audience is that we just keep coming back to this. AGI is a really interesting topic; it's fascinating to sort of follow along progress towards it. It's a meaningless term related to what it does to impact your job, your company, the economy more broadly. We don't need to reach AGI—whatever that definition is—we don't need to agree on a definition for AI to transform businesses, the economy, and society. This idea of capabilities overhang—we talk about that. That Andrej Karpathy episode that I mentioned touches on this quite a bit. But just go back to that example I shared of rocks. If you have a company like ours that knows this stuff, we understand what AI capabilities are and we look at the operating system of our company and we're like, "Oh, we're just going to reimagine the whole thing rather than five rocks a quarter, we think we can do 15 or 20 easily." And here's how we're going to do it. We understand the capabilities and we're applying them to the best of our ability.

Then take some other company that doesn't even have GenAI tools for their team yet. They don't even have Copilot licenses or ChatGPT licenses. They've done no personalized training, they've never run a workshop internally—they're not even taking advantage of any of the capabilities other than maybe using it as an answer engine or a chatbot. There's this overhang of all these capabilities and so few companies are actually doing anything with them. Not just companies—educational institutions, governments, practitioners at an individual level. So that to me is the most important thing. I'm all for this—I think quantifying it so we can just get to the point we agree on what it is makes total sense, but don't be misled by that or wait around for that definition to be like, "Oh, okay, I'll worry about it when we get closer to AGI." It's already there.

What 81,000 People Want from AI

[Mike Kaput]: All right, next step Anthropic has published results from the largest multilingual qualitative study ever performed on AI attitudes. They did nearly 81,000 interviews with Claude users across 159 countries and 70 languages. These conversations were actually conducted by Anthropic Interviewer, a variant of Claude trained specifically to conduct and then analyze interviews, which we have talked about in past episodes. Interestingly, they found out that the top fear people expressed in these conversations is actually hallucination and unreliability of AI, which ranks as the number one concern with 26.7% of people mentioning it. It is ahead of jobs and economic impact, which is at 22.3%, and loss of human autonomy and agency at 21.9%.

Interestingly, Anthropic finds that people value AI often for the same capabilities that they fear most. 50% of respondents experience time savings from AI, yet 19% felt pressured to simply work faster as a result. 33% cited learning benefits, while 17% worried that it would actually facilitate more cognitive decline when you're relying on machines to think for you. And it's interesting that people experiencing one side of a tension are typically three times more likely to also worry about the other side, meaning these are kind of inherent contradictions in the same people using AI. What's really cool is they actually asked what people actually want from AI. 18.8% of those who answered said that they seek professional excellence from AI. 13.7% said they were seeking personal transformation. 13.5% said better life management, and 81% report experiencing some progress towards their vision in those areas. Paul, I'm interested what you took away from this data. Pretty interesting way they went about getting it.

[Paul Roetzer]: The approach to research is what I found most intriguing. The data is great. I do think again, as I referenced earlier, you have to keep in mind who are the people responding to these questions when you look at the data, so you're not making some broad assumptions. In December of 2025, before Claude Code really took off and before the government issues and before this movement toward the Claude app became the number one app on the App Store, they have a heavy technical user base. Lots of coders, lots of AI researchers using Claude. So when you're looking at this, even though it's 80,000 people across all these countries, it's still likely skewed toward a more technical user. Just for reference sake, that's important to keep in the back of your mind.

I love the approach—this dynamic approach based on responses that adapts it. Not great news for people who run focus groups and who are consumer research people for a living. This is definitely one of those ones where you're either adapting or the whole new way of doing research is going to run you over. They said their next Anthropic Interviewer study launching shortly to a small subset of Claude users focuses on Claude's effects on people's well-being over time, whether Claude is actually making people's lives better in the ways they want and how it could do so more effectively, which I thought was interesting. They said this is a new form of social science that is qualitative research at a massive scale, and they're in the early stages of learning how to do it.

Surveys and usage analysis tell us what people are doing with AI, but the open-ended interview format helps us get at the why. Conducting this research has moved them and challenged them. They did not expect so many deep, open, and thoughtful responses. By far, the most common reflection from their team was that it was viscerally moving to see Claude impacting people's lives for the better, and equally motivating to hear their concerns. They were equally gripped by the fears and downsides—people saying that the same availability making Claude useful is what makes it hard to put down, or knowledge workers worrying about outrunning AI's economic impact. When you come into contact with this much raw human experience, it knocks you sideways, they said. The usefulness is real, and the question for all of us is how to claim the benefits without incurring undue costs.

I thought that was really interesting to note, Mike, because this actually came up during our company retreat—this idea that we're all sort of at the frontiers of figuring all this out and using it, and it's awesome for productivity and innovation and efficiency and growth and all these things. But it also has this very messy, complicated other side where it has this human impact. Maybe your friends or your family hate it, and they don't even like the fact that you're working on it, and they have these perceptions about what you're doing because you're in AI or because you're one of the people who talks about AI. I honestly think about that sometimes from what we do on the podcast, Mike. I think, "God, I hope at some point people don't—we're trying to do the human-centered work. We're trying to educate people so we can have a positive outcome." But sometimes the truth doesn't matter, and I do worry about that. It's part of the reason I don't read comments on social ever. I don't look at our comments on YouTube and X and maybe sometimes LinkedIn, but I just prefer to try and just do our thing and know we're trying to do a positive thing, but that doesn't change the fact that there's darkness to this, and there's uncertainty and fear and anxiety and hatred—all those things are very real. I'm really excited actually that Anthropic's going this research direction.

[Mike Kaput]: That's why I really actually like the findings here. Obviously, to your point, they are skewed towards a certain type of people. But yeah, when someone asked at our offsite how you stay grounded when you're dealing with such heavy and sometimes horrible dark AI topics in the news, that was my answer—focusing not to the detriment of the negative, but focusing on the positive things I've been able to do with these tools. I've been able to do things, achieve goals, get results that I never dreamed possible. This technology has made me a better professional, leader, thinker, strategist, even husband and father. That's kind of the flip side. I love to see in this data people seeking professional excellence or personal transformation or better life management. I've done all those things with AI, and it is glorious what you're able to do.

[Paul Roetzer]: It doesn't get rid of the negative stuff or the concerns, but it's trying to focus on the positive. We brought on our director of research a few months back, and it's one of the focus areas she has—the humanity side of this. Mike and I and Taylor are actively talking about more research in these directions and the kinds of things around the human impact. It's something we're going to probably be doing a lot more about on the show and then even with our academy. And maybe even on our event side, we might be looking at doing some stuff where we can bring people together to have these conversations because they're critically important.

AI Academy Spotlight

[Paul Roetzer]: As we wind down, Mike, we had mentioned at the start the AI for professional services, which you taught as part of our academy. One of the ideas we have is to do little spotlights on these where we give you a little bit of insights into some of the key things we learned in building these courses. Mike, your AI for professional services—any key insights or takeaways that you think would be helpful for people to hear?

[Mike Kaput]: Sure, Paul. As part of this four-course series, which comes with its own certification, we're breaking down both from a high level what is happening at the industry level that you need to know about, and then getting into the actual tactical A to Z of how you identify your own use cases and match AI tools to them in your own professional services career. A couple things that just jumped out as part of both building this course and as someone that was in professional services before we did the whole AI thing.

Number one—and we've talked about this on the podcast—one of these trends that really needs to be appreciated is the idea that the billable hour model is maybe not only on borrowed time, but is dead. If you were on a billable hour model as a professional services organization, AI is a major threat to that because many organizations still have not adequately figured out what happens when you can now do things in a fraction of the time that you used to do them in using AI. You cannot simply charge the same amount of hours and hope to get away with it. You see a lot of industry professionals and leaders trying to figure out how we adapt our business model without tanking our entire organization. One of the big takeaways there is the firms that are going to win are the ones that figure out sustainable, defensible, value-based pricing first—pricing on outcomes, not hours. Because again, you can do so much more in the same amount of time. There's no chance your clients and your customers are not going to demand that you pass along those savings to them.

I would also say another big area here is figuring out how your human intelligence within your professional services firm becomes your superpower and your competitive advantage. Because, unfortunately, for a lot of professional services firms, there are very intelligent AI models out there that now have been, for better or worse, trained on a lot of your expertise. Figuring out how your humans, with all their experience and background and domain expertise, can actually be leveraged and scaled with AI is going to be the entire battle moving forward. You really want to look at any frameworks, any experience you have internally as almost like your own IP if you're not already, because AI can scale that, and you can have that be a competitive advantage. But if you do not do that, if you are playing at the commodity level of, "Hey, we're experts in marketing," so is AI now. You have to figure out what kind of expert you are and how you are differentiated.

Last but not least, there are always these questions in professional services about how we'd love to get started with AI, but we work in really sensitive industries with clients that have privacy and data concerns about using this stuff. Totally valid. We talk about that more at length in this course series, but the advice here is actually to start with your back office stuff. If you have these kinds of challenges, if you are still trying to navigate data and privacy concerns, your back office stuff, I guarantee you, can become dramatically more productive by applying AI, often at a very low-hanging fruit type of level. We go into very specific use cases and tools in the course series to help you do that. But there are these areas where they don't touch client-facing stuff that you can actually start your AI journey almost in the back office and achieve massive immediate profitability gains just from doing that alone.

[Paul Roetzer]: The other thing I think about, Mike, is just from the buyer perspective—understanding the professional services and how it's evolving and how I should be looking for AI-forward professional services firms. Even for me as the CEO, we outsource legal, IT, accounting, advertising. I just think about those four. Understanding how their business models are evolving and the importance of working with AI-forward versions of those companies and the points of contact, things like that. It is great and I appreciate you building the series and this ongoing effort we're doing to try and sort of create content across all the departments, all the relevant industries, and then even into businesses. Hopefully these little spotlights will be helpful for people to get a little taste of what's going on in these different industries. We'll touch on departments, we'll touch on some of the GenAI things we're doing, and just try and bring some of that value from Academy to the podcast each week.

AI Product and Funding Updates

[Mike Kaput]: All right, Paul, we've got a number of AI product and funding updates here to wrap up this week. First up, Jeff Bezos is trying to raise a hundred billion-dollar fund focused specifically on AI manufacturing. This fund would represent one of the largest single pools of capital ever assembled around AI infrastructure.

Google has launched something called Stitch, an AI design tool that turns natural language prompts into high-fidelity UI design. The tool lets you describe what you want in plain English and generate production quality design output. Google is kind of in this emerging "vibe design" category. Google also rebuilt AI Studio from scratch as a full-stack vibe coding platform. They said they spent four months on this rebuild and the new version lets developers go from prompt to working application entirely within AI Studio.

OpenAI has released smaller, cheaper tiers of GPT-5.4. GPT-5.4 mini and nano give developers access to the model family at lower cost and latency.

In some other legal news, a court temporarily allowed Perplexity's AI shopping agents to continue operating on Amazon. Perplexity's agents browse Amazon on behalf of users to actually find and purchase products and this ruling lets the service remain live while their ongoing legal dispute with Amazon plays out.

On X, the company is rolling out AI-generated article summaries that appear when users share links on the platform. Researcher Ethan Mollick noted the irony that many of the articles being summarized are themselves obviously AI-generated. We're creating an interesting loop where AI summarizes AI.

[Paul Roetzer]: And then it trains the Grok language model. Part of the reason why they made articles such a prominent feature is to get a lot more training data that was probably trained with them potentially.

[Mike Kaput]: Finally, and we'll be keeping a close eye on this one, Demis Hassabis, CEO of Google DeepMind and Nobel Prize winner, is teasing his upcoming book called The Infinite Machine, set for release on March 31st. It covers the story of DeepMind and Hassabis's vision for the future of AI. I'll be looking very closely at that one, Paul.

[Paul Roetzer]: I did pre-order this one. This is a good way to end today's podcast. I'm actually going to read the excerpt because I think this is really fascinating. This comes from The Infinite Machine:

"The true reason to build artificial intelligence, Hassabis was now saying, went beyond Kant and Feynman. The goal was to draw closer to what might be called God, to the intelligence that may presumably have designed everything around us. Hassabis quote: 'I am first and foremost a scientist. My goal is to understand nature. But doing science is sort of like reading the mind of God. Understanding the deep mystery of the universe is my religion, kind of. We humans, we have these faculties. The world is understandable, but why should it be that way? I think there is a reason. Computers are just bits of sand and copper,' Hassabis continued, now sounding more urgent. 'Why should these combine to do anything? I mean, it's absurd. The electrons move around and then that creates an AI system that can defeat a Go master. Why should that be possible? This table,' Hassabis rapped his palm on it for emphasis, 'why should it be solid? This is beyond evolutionary coincidence. We can build electron microscopes and interrogate reality down to the most minute detail. We can build systems that detect black holes colliding more than a billion years ago. I mean, what is this? What the hell is going on here?' There was a pause, but Hassabis was not yet finished. 'I sit at my desk at 2:00 a.m. and I feel like reality is staring at me, screaming at me. Literally screaming at me, trying to tell me something if I could just listen hard enough. That's how I feel every day. So, you can see why I'm trying to build AI. I felt that since I was very young, that there's a deep, deep mystery about what's going on here. You can frame it how you want. You can call this God's design or you can say it's just nature. I'm open-minded about the description and I don't know what the answers will turn out to be. But at the moment, we don't really know what time is or gravity is or any of these things. So, there's a mystery waiting to be solved and it encompasses just about everything. I would like to understand before I croak. I would like to understand and then I'm perfectly fine to shuffle off my mortal coil.'"

That was awesome. Demis thinks very deeply about this. Elon actually commented on that one. He said, "I share Demis's urgency and thoughts here." I think it's important to understand why one of the people is building AI. And it is for a much bigger solve—solve intelligence and then solve everything else. That's been his mission for the last 30-plus years of his life.

[Mike Kaput]: Go to smarterx.ai/pulse to take this week's survey. We're going to ask a couple questions about the topics this week. One is about OpenAI's enterprise deployment with that private equity backing we discussed. The second one is about Anthropic's study and some of the findings there and how you feel about them. We'd love to hear from you. Paul, really appreciate you breaking down everything for us this week.

[Paul Roetzer]: Good stuff. Busy week as always. I think we just have one episode this week. I have to check my calendar yet. Maybe we have a second one, but we'll be back next week and then I think I'll be on spring break then for like 10 days. So, my next week might be—we might be on a break after next week. Thanks for being with us. Have a great week, everyone, and we'll be back with you next week.

Key Takeaways

• The Agent Race is On: Major AI labs (OpenAI, Anthropic, Google, Meta) are shifting focus from general chatbots to autonomous agents and enterprise deployment.
• Claude Code’s Impact: Anthropic’s Claude Code has set a new bar for agentic capabilities, forcing competitors like OpenAI to deprioritize "side quests" and double down on coding and research agents.
• Political Polarization: AI is rapidly rising as a political issue. Polling shows high voter concern regarding job losses and a lack of government planning, while industry super PACs push for continued acceleration to maintain American dominance.
• Company Transformation: AI is drastically compressing the time required to complete complex business tasks. Projects that previously took months can now be prototyped in minutes using advanced models like Claude 4.6 Sonnet.
• The Future of Professional Services: The billable hour model is under threat. Firms must pivot toward value-based pricing and leverage AI for back-office efficiency to remain competitive.