The AI Daily Brief: Artificial Intelligence News and Analysis - Why Your Company Needs to Move Faster on AI

Episode Date: June 13, 2025

NLW argues that 1) things are moving faster than we think; 2) AI will be even more transformative than we think; and 3) that the winners of a decade from now will be made in the next 1-2 years. Find out why. Get Ad Free AI Daily Brief: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://patreon.com/AIDailyBrief⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Brought to you by:KPMG – Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://kpmg.com/ai⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ to learn more about how KPMG can help you drive value with our AI solutions.Blitzy.com - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://blitzy.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ to build enterprise software in days, not months AGNTCY - The AGNTCY is an open-source collective dedicated to building the Internet of Agents, enabling AI agents to communicate and collaborate seamlessly across frameworks. Join a community of engineers focused on high-quality multi-agent software and support the initiative at ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠agntcy.org ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Vanta - Simplify compliance - ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://vanta.com/nlw⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Plumb - The automation platform for AI experts and consultants ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://useplumb.com/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠The Agent Readiness Audit from Superintelligent - Go to ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://besuper.ai/ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠to request your company's agent readiness score.The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast version of The AI Daily Brief wherever you listen: https://pod.link/1680633614Subscribe to the newsletter: https://aidailybrief.beehiiv.com/Join our Discord: https://bit.ly/aibreakdownInterested in sponsoring the show? [e-mail]

Today on the AI Daily Brief, why your company needs to move faster on AI. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Hello, friends, quick announcements before we dive in today. First of all, thank you to today's sponsors, Blitzy, Plum, Vanta, and Super Intelligent. And to get an ad-free version of the show, go to patreon.com slash AI Daily Brief. Now, for today's show, we are doing something a little bit different. This is my last day of travel, and we've mostly been able to keep a pretty standard format, but today I am doing a shortened version of a presentation I gave recently about why companies

need to think in even bigger ways than they are currently thinking now. The presentation is called why your company needs to move faster on AI, and I will be doing an abbreviated version here, although abbreviated still means 39 slides. For some of you, this will be well-trodden territory, although there may be some new facts and figures that you haven't heard before, but really the audience for this is whoever you're trying to convince at work needs to see the world the way that you do. So if you are looking for an episode to get them on board with your big push to move AI and agents faster in your company, send them this episode. So here's the TLDR and the big idea. First of all, I'm going to argue that things are moving even faster than they seem, that the disruption is going to be even bigger than it seems,

and that the winners of the next five to 10 years will be built in the next one to two years. It is the standard convention of presentations and conversations like this to hedge a little bit, right? and to say, yes, AI is powerful, but it has to be deployed responsibly. And here are some considerations around where it's not really all that good yet. I will be doing exactly none of that. There are plenty of places out there that will tell you to hedge or slow down or be considerate. I am here to tell you that I believe that the cost of underinvesting in AI is significantly higher than the cost of overinvesting in AI.

If you overinvest, chances are that you spend a bunch of money building out something that's going to be obviated by technology advances in just a few months. Perhaps in the interim, you will overspend on technology that becomes a lot cheaper. Those are real costs, certainly. The cost, I believe, of underinvestment is actual organizational extinction. If you do not adapt, your competitors will, and you will not be in a good position on the other side of that. But let's talk first about this idea that things are moving even faster than they seem. Going back to when Chat Chachapit launched, it was the fastest platform ever to reach 100 million users.

Previously, the fastest had been TikTok at about eight months, and Chat Chachapit took five weeks. Overall, Gen A.I. Adoption, according to the Federal Reserve Bank of St. Louis, is about twice as fast as Internet adoption. What this chart shows is that whereas it took the Internet about five years to reach 40% penetration in U.S. households, that only took Gen A.I. about two years. Growth has not just been in the consumer's theory. It has, of course, been in the enterprise as well. This is a woefully out-of-day chart for McKinsey that shows that we went from zero companies using Gen A.I. before it existed to now it becoming fairly ubiquitous. Importantly, the uses are all over the enterprise. They are not just in one category. Gen AI is being used in everything from data analysis to customer support to task automation, to content generation, and beyond. This is a chart from a recent writer survey that talked to 800 executives and 800 employees, and one of the things that you'll see here, which we will come back to later,

is that there's a fairly big gap between how the C-suite is using AI and agents and how general employees are. This is one area where the C-suite is in general in enterprises out ahead of their employees, and that's actually causing some problems. Still, the biggest thing on top of all of this is not only that it's moving fast, but that it's accelerating. These massive Silicon Valley mouthwatering up into the right charts

are from OpenAI's users, subscribers, and revenue, and you can see how even though it was fast out of the gate, it has gotten nothing but faster. It took about two years for ChatGPT to reach 400 million weekly active users, and then just about two months to double to 800 million. Acceleration is once again happening in the enterprise as well. This is a chart that shows growth in organizing, organizations that are using AI for one or more functions, two or more functions, three or more functions,

you can see a very similar pattern. Between 22 and 23, there was growth, although modest, as organizations wrap their head around chat GPT and all the new things coming with it. But then in the first half of 24, you saw a big leap. For example, the percentage of organizations that were using AI for two or more functions jumped from 31 to 50% in the first half of 2024. In many cases, there was an even steeper jump in the second half of 24. For example, the number of organizations using AI in three or more functions jumped from 27 to 45% in the second half. My guess is that when we see the update of this chart, it will be even more up into the right. Now, it's accelerating in part because everything in AI is compounding. One important area of that is, of course, in the developer ecosystems. These charts are

from Mary Meeker's recent AI trends report and shows a big 6x growth in the Nvidia developer ecosystem between 2019 and now, and full 5x growth in the Gemini ecosystem between May 24 and May 25. The point here is, of course, that the more developers, we have building things in AI, the more things in AI that people have access to, the more people use AI, and so on and so forth. Importantly, and if you are a regular listener to the show, you will know that this isn't just traditional coders and software engineers who are getting their hands dirty with code. One of the most important trends of the moment is, of course, vibe coding. You can see here that interest in vibe coding has started to exceed prompt engineering,

that it's rapidly accelerating in terms of the number of repositories on GitHub. And the reason that vibe coding is so important is that it's opening up entire new categories of use cases. When non-developers can speak in code to developers, not only can they interact and move their organizations forward faster, but they unlock AI use cases that simply weren't possible before. They're no longer constrained in the same way by whatever startups happen to have put out when it comes to AI applications. This is a force in the market that is only just starting and I believe is going to have a dramatic impact that will further accelerate this whole movement. Another key area of improvement has, of course, been in the models themselves.

The shift from non-reasoning to reasoning models has opened up in a fundamental way this new agentic era. Speaking of agents, if there is any word that competes with AI for most buzz and most hype, it has to be agents. Agents are important in the enterprise for a fundamental reason. They change the way that enterprises think about AI. In the co-pilot and assistant era, it was very easy for organizations to think about AI strictly as a productivity enhancement technology. So basically, a technology that helps their existing employees do the things that they do now, but a little faster, a little better, or a little cheaper, or some combination thereof. Productivity enhancement technologies are great.

The power of doing everything that you do now, but 30% faster, cheaper, or better, is enormous. Entire sectors could be rearranged on the basis of that sort of productivity enhancement. However, that is very clearly not the big game when it comes to the way that AI will transform industries. With agents, you have the opportunity to fundamentally rethink business models, deployment models, operational models, in ways that are about way more than just productivity. And organizations and enterprises get this. As they start to think about agents, they are no longer just in that paradigm of efficiency, but thinking about growth and opportunity. Now, interest in agents is showing up in the numbers as well. This is from the KPMG Pulse survey

in between Q4, 2024, and Q1 of 2025, the percentage of enterprises piloting agents nearly doubled from 37% at the end of last year to 65%, almost exactly two-thirds of organizations piloting in the first quarter of this year. What's more, and I think this is even more important? The sentiment and vibe, frankly, around the agent pilots is very different than a year ago's sort of co-pilot pilots. Whereas before this was a question of if, an exploration of how good this technology was and what it might do inside the organization, even though agents are incredibly nascent now and very limited in their functionality, effectively no enterprises that are piloting them are thinking about them as an if. Instead, they are being viewed as completely inevitable. There is a presumption of future

transformation that will come from them, and the pilots now are just getting in reps to figure out what's actually available and what organizational structures need to be rebuilt to accommodate them. You see this once again in the KPMG survey that found 99% of respondents were planning to deploy AI agents. And frankly, I have to assume that the other 1% misread the question. Now, this intention that I was just mentioning is transformative. It's very different when your organization is behaving as though it's 100% sure that a new technology will upend how it does things. And you're seeing that intention show up in the budget. According to PWC, more than 80% of organizations saw AI budget increases this year due to agents. And a little more than two-thirds of those

saw 10% or more budget increases. And as I said, even though agents aren't necessarily necessarily all the way there yet. Things are changing quickly. Research group METR found that agent capabilities have been doubling about every seven months. Their methodology for this was to see how long a task an agent could successfully do at a success rate of 50%. And then while there's tons of good room for debate around whether that's good methodology or not, it is at least very consistent methodology. And so even if you believe that there might be a better way to determine this, what matters isn't so much your determination of whether that's a good metric of agent success, but simply how the trend line is improving. What's more, it appears that the rate is increasing now,

and that agents are doubling in capacity closer to every 70 days. What's more, it's not just model advances that are increasing the acceleration in agents. We also have very, very fast alignment around standards that's accelerating the field as well. If you know any initials associated with agents, they're probably MCP, which stands, of course, for Model Context Protocol. Model Context Protocol is a way to give agents easy access to data sets. So you can build an MCP server that connects to a particular data set, and then other agents can plug into that MCP server in a way that is much simpler than having to wire their agent to that data in a more manual or bespoke way. Anthropic introduced MCP in November. It had some initial uptick in excitement, and then starting in about February,

it really took off and has never looked back. Subsequent to that, every one of the big labs, despite the fact that they are all anthropic competitors, has all done. jumped on the MCP train. Google has adopted it. OpenAI has adopted it, even though it looked for a time like they were going to try to compete on this standard front, and Microsoft has adopted it. In fact, Microsoft has not only adopted MCP and made it a key feature of Microsoft build this year, but also has their CEO, Satya Nadella, tweeting about A2A, which is a messaging protocol for agents that comes from Google. And this is a historical behavior when it comes to standards. The protocol wars around core internet standards, like for things such as email, were more

multi-decade battles. Very clearly when it comes to AI and agents, all of the big labs have decided that the additional value and speed they get from everyone aligning around the same standards is worth way more than being the company that built those standards in the first place. The implication of all of this is we cannot design for where agents are now. We have to design for where they're heading. And on that front, Microsoft is starting to lay out a vision for what the future of organizations looks like when they are agent-enabled. These year's Work Trend Index was all about what they call the frontier firm. They see the frontier firm happening in three phases. Phase one, which is roughly where we are, or at least where we've been, is human plus assistant. Every employee

has an AI assistant that helps them work better and faster, right? This is the efficiency and productivity era. Phase two, where we were just starting to get into now is human agent teams. Agents join teams as digital colleagues taking on specific tasks at human direction. This is what a lot of the next six to 12 months is going to be about. Phase three, they call human-led agent-operated. human-set direction and agents execute business processes and workflows checking in is needed. The paradigm for them is called agent bosses, where everyone, instead of doing the work that they used to do, manages agents that do that work for them. But what does this look like in practice? Well, OpenAI recently published a paper to try to help enterprises understand what use cases

they might think about, and they organize the world into six use case primitives that they saw most commonly across all of their organizational and enterprise deployments. The six simple AI use case categories were content creation, research, coding, data analysis, ideation and strategy, and automation. And so trying to apply this phasing and sequencing of how agents might impact something like content generation. Today, we have LLMs that can draft copy, translate things, and repurpose formats. So if you've made a YouTube video, you can rip that transcript out and turn it into a blog post for LinkedIn. That's incredibly valuable, right? Serious efficiency gains and new opportunities come from that. But over the next six to 12 months, or whatever

the period is, over the next phase, let's call it, you're going to start to see the introduction of background agents as well. These will be things like context-pinned ghostwriters and agents that monitor brand voice, campaign goals, etc. And then auto-generate documents post-video scripts that are routed to humans for approval. A phase after that will see even more advanced background agents, doing things like watching engagement metrics in real-time, running AV tests, and revising creative content continuously. Oh, and even further phase out sees full synthetic creative studios where you have multi-agent teams, think agents that are writers, agents that are designers, agents that are voice actors, etc., who can storyboard, shoot, edit, localize, and place

ads end-to-end. This episode is brought to you by Blitzy. Now, I talk to a lot of technical and business leaders who are eager to implement cutting-edge AI, but instead of building competitive modes, their best engineers are stuck modernizing ancient codebases or updating frameworks just to keep the lights on. These projects, like migrating Java 17 to Java 21, often means staffing a team for a year. year or more. And sure, co-pilots help, but we all know they hit context limits fast, especially on large legacy systems. Blitzy flips the script. Instead of engineers doing 80% of the work,

Blitzy's autonomous platform handles the heavy lifting, processing millions of lines of code and making 80% of the required changes automatically. One major financial firm used Blitzy to modernize a 20 million line Java code base in just three and a half months, cutting 30,000 engineering hours and accelerating their entire roadmap. Email Jack at blitzie.com with Modernize in the subject line for prioritized onboarding. Visit blitzie.com today before your competitors do. Today's episode is brought to you by Plum. If you're building agentic workflows for clients or colleagues, it's time to take another

look at Plum. Plum is where AI experts create, deploy, manage, and monetize complex automations, with features like one-click updates that reach all your subscribers, user-level variables for personalization, and the ability to protect your prompts and workflow IP, it's the best place to grow your AI automation practice. Serve twice the clients in half the time with Plum. Sign up today at Useplum.com. That's U.S.E.PLUMB.com forward slash NLW. Today's episode is brought to you by Vanta. In today's business landscape, businesses can't just claim security, they have to prove it. Achieving compliance with a framework like SOC 2, ISO-27-01, HIPAA, GDPR, and more is how businesses can demonstrate strong security practices.

The problem is that navigating security and compliance is time-consuming. and complicated. It can take months of work and use up valuable time and resources. Vanta makes it easy and faster by automating compliance across 35 plus frameworks. It gets you audit ready in weeks instead of months and saves you up to 85% of associated costs. In fact, a recent IDC White Paper found that Vanta customers achieve $535,000 per year in benefits, and the platform pays for itself in just three months. The proof is in the numbers. More than 10,000 global companies trust Vanta. For a limited time, listeners get $1,000 off at Vanta.com slash NLW. That's V-A-N-T-A-D-C-N-L-W for $1,000 off. Today's episode is brought to you by Superintelligence, specifically Agent Readiness Audits.

Everyone is trying to figure out what agent use cases are going to be most impactful for their business, and the Agent Readiness Audit is the fastest and best way to do that. We use voice agents to interview your leadership and team, and process all of that information to provide an Agent Readiness score, a set of insights around that score, and a set of highly actionable recommendations on both organizational gaps and high-value agent use cases that you should pursue. Once you've figured out the right use cases, you can use our marketplace to find the right vendors and partners. And what it all adds up to is a faster, better agent strategy. Check it out at besuper.aI or email agents at besuper.a.i to learn more.

Now, this came out of a collaboration between me and OpenAI's O3 model, applying these Microsoft phases to the OpenAI use case primitives. But I actually think that this still undersells how big the agentic change is going to be. If you are a regular listener here, you will have heard me squawk on about the Dr. Strange Theory of AI agent work. But again, hopefully some of you are hearing this for the first time. And for those of you who aren't familiar with the reference, spoiler alert from something that came out about a decade ago now, but in the infinity cycle, which was Marvel's big.

event in the teens, the Avengers are getting prepared to fight Thanos, who is the big bad of the series, who wants to eliminate half of the universe with a snap of his fingers, and things aren't looking good for them. To try to find any source of intelligence or inspiration, Dr. Strange goes into the multiverse to review all of the multiversal scenarios in which the Avengers are fighting Thanos to see where they actually succeed. After reviewing 14,605 scenarios, he found there was only one in which they beat the Mad Titan. Now, that sets up the entire context of the last two movies, but what I think is interesting about this is that in a world of unbounded intelligence, that is to use a phrase that has been popularized by Sam Altman, too cheap to meter,

why can't we bring a similar way of thinking to work? In other words, why can't we, for every decision, review 14,605 scenarios? So what would that look like in the same context of content generation? Well, let's imagine social media copywriting. Right now, even the most sophisticated organizations, and even 03 when I was talking to it, are still basically seeing a one-to-one replacement or relationship between the work that humans do now and the work that agents will do in the future. And so in the case of copywriting, you might have a small team of copywriters that focus on different channels. But instead of having one copywriting agent, why not have a hundred, each having a different voice? Maybe some of those are meant to mimic

your brand voice. Others are meant to mimic a competitor's brand voice. Others for fun are meant to mimic famous writers like Shakespeare or Hemingway. From there, you have another set of agents who are trained to imitate key audiences and who provide reviews of the different writing agent versions based on those audiences preferences. Finally, you have a set of synthesizer agents who take in results and put it together all into a short list of options with reasoning, making recommendations to humans who can finalize. Now, on the one hand, there will certainly be work contexts where this is total overkill. Not every decision will justify this mode of working. But the constraint will not be in our ability to apply this sort of intelligence to any given problem, even something

as small as a tweet. We just got last week an 80% drop in the cost of O3, which was until that same day, OpenAI's most advanced model. 80% drop in three months since O3 had been launched. And by the way, when O3 came out, it was already cheaper base than O1 had been just a few months before that. Cost of intelligence is coming down precipitously, and as it does so, this is the sort of scenario that's going to be possible. We won't just see a one-to-one replacement for today's human workers. We'll see combinations of intelligence and work on a magnitude and scale that were never before possible to even contemplate. So the big question is, of course, how are winning organizations preparing for change of this magnitude? And the short, unfun answer if you are in those

organizations and leading this charge is you got to do it all at once. I see six areas that I think every enterprise needs to be thinking in terms of. First of all, there has to be leadership frame setting. You might remember that chart that we looked at at the beginning that saw that C-suite executives were doing more of basically all the use cases than employees. And we very frequently at superintelligence see a big gap in leadership between where executives think they are with their AI strategy and where employees think that they are. This, by the way, is replicated every time people survey about it. That recent writer survey from December of last year found wildly different sensibilities between executives and their employees about things like, do I have an AI strategy?

I think the respondents to that were something like 75% of executives said that they had an AI strategy versus just 45% of employees said that they had an AI strategy. And the stakes of the leadership conversation are not just informational. When employees don't have a sense of the leadership vision of, especially agents, they can have real concerns around where they're going to fit in the organization in the future. Those are not technical questions. Those are leadership questions. and leadership has to engage with employees, not just to tell them that AI is important, but to help them understand how AI is going to impact the shape of the organization in the future. So leadership is one thing you've got to be doing.

A second thing you have to be doing is that even though we are kind of moving on to agents, the individual productivity gains that come from co-pilots and assistants are still really important. Organizations need to be investing in the bottom-up capabilities of their people to use these tools. And I think of anything we've observed, there's been sort of an overcorrection to just focus on agents, and a lot of those things that people cared about last year, i.e. upskilling and training programs, have kind of been kicked back to an L&D function that tends to get organizational second shrift. I do think it's very important for organizations to be investing in the bottom-up capabilities of their people to use these tools, and so that's a second area.

Third, yes, in fact, your enterprise does need to be experimenting with agents. And it's okay if you realize that agents aren't at a point where they can be fully deployed for your mission-critical functions. Smart organizations are saying, fine, we are still going to to go test things in other areas. We're going to test SDR agents, research agents, customer service agents. Basically, they're not using the excuse that agents aren't exactly where they're going to be in the future to not start today. Fourth, one of the most important categories right now is data and technical infrastructure. There are huge changes happening to the way that agents are being built, all of which point to the supremacy and importance of having a sophisticated data architecture

that can interact with whatever end models we end up using. Likewise, there needs to be be a policy infrastructure investment. People are using AI in a lot of inappropriate ways at work, in many cases not because they're trying to break the rules, but because they don't know what the rules are. Policy infrastructure is ultimately a key part of things right alongside data infrastructure. And again, the best organizations are thinking about this all at once. Lastly, and really importantly, there has to be a future visioning process, a future reimagining again, the best organizations are not just thinking about AI as an efficiency and productivity tool, but really trying to understand how it's going to shape their business model, their

deployment model, their operational model, basically what they do and how they do it on a fundamental level. Now, as I just said, you kind of got to do all of this at once. However, if you were going to spend the next six months doing just one thing, there's a lot to be said for really investing in the technical infrastructure. According to the economist, only 22% of organizations right now say that their current architectures are fully capable of supporting AI workloads, and there is just a ton of work to get that up to speed. Now, a key trend that we're observing at super-intelligent is that while in many cases organizations will, yes, use sort of off-the-shelf agents,

for many high-value and highly regulated industries, they're going to be forced to roll their own AI systems. Luckily for them, there is a huge amount of work being done right now around agents that effectively build other agents. Inside those types of organizations, that might become the norm, where everything is about the technical and data infrastructure that models can be dropped into for a really bespoke and customized experience. But that's only going to work if that data and technical infrastructure is up to snuff.

So what are some common traps for organizations? One is waiting for future model improvements to start. You see this all the time. There's a sort of paralysis that happens because you know that things are going to be better six months down the line. So how much should you invest today? Well, I will tell you as someone who runs a startup that has spent six months building, tweaking and tuning a voice agent. That I'm almost positive in six months from now, I could have vibe coded in the day. You just don't get the privilege of waiting. A second trap is treating this as

though it were just a software change. This is not like shifting cloud platforms. This is fundamental reimagining of how the organization is going to work and you have to treat it as such. Next, we've got overindexing on efficiency and cost mindset, on process automation rather than business redesign. There is a tendency for some organizations to view agents as just RPA2.0. And I think that those organizations are going to get smacked, which of course is not to say that you shouldn't be thinking about automating processes, but that is far from the be-all-end-all. As we just discussed, another trap is not setting a leadership vision and not bringing employees into that vision, and overall, there's just a dearth of thinking big enough.

Now, I can leave you with one very solvable problem relative to many of these others, which are much harder. Another key challenge that we see is bad tools. There tends to be a fairly big gap between the quality of the consumer tools that people are using in their regular lives, i.e. ChatGPT in its most advanced models, and the crappy a couple generation ago models that they have through work. If you've gone from using O3 in your personal life to using what Microsoft co-pilot consists of in most organizations right now, it's a very frustrating and limiting experience. Same when it comes to coding tools. Obviously,

this is an area that is one of the most fast moving, and you're actually seeing some big organizations start to ease up on this. Most recently, it looks like an internal revolt has Amazon considering ditching its own AI coding assistant to instead use cursor. I call this one out specifically because, like I said, in a world of very complex challenges, this one isn't all that hard. Figure out how to get your legal and compliance on board and use better tools. I promise you he will be glad that you did. Ultimately, like I said, if you have to take away one thing from this, it's that AI is not just about efficiency. It's about opportunity. And rewiring your mindset to think in that way, and then rewiring the organization to seize that is the key. The winners of five to ten,

10 years from now will be made in the next one to two years for a very simple reason. Agent and AI Advantage compounds. If you sit back and think to yourself, I will just get into this game in six months when agents are better. What's going to happen is not just that you're going to be six months behind your competitors. It's that they are going to go through reps of discovering all the organizational change that needs to happen to fully utilize these tools. They're going to go through a data readiness process.

They're going to develop new policy infrastructure. And what that means is that when new model capabilities come online, even if you are now in the game when they do, you will not be nearly as ready as they will to take advantage of them. And as you are then racing to get your organization up to snuff, not only will they be using those advanced tools, but those advanced tools will be allowing them to move faster than you, so the gap between you and them will be increasing, not decreasing. You get the gist here. The point is that agent advantage compounds, so you've got to move faster. That's my spiel for today, guys.

Like I said, hopefully this was fun for you if you are familiar with this, and hopefully this is a useful tool for you to win arguments internally. I would love to hear stories of you slapping this down on someone's desk to finally get the resources you need to push the organization in the direction you needed to go. For now, I appreciate you guys listening and watching, as always, and until next time, peace.