275 | Knowledge work as we know it is over, self improving agents loops connected to Microsoft or Google eco systems are now possible, build entire software suite autonomously, Meta is cutting 20% of it’s workforce, & more AI news for Mar 13, 2026

What happens when AI stops helping with work—and starts doing the work itself?

This episode connects a set of developments that business leaders should not ignore. From Sequoia’s thesis that AI is replacing services, not just software, to Anthropic’s findings that AI adoption is still far behind AI capability, the message is clear: the bottleneck is no longer technology. It is implementation.

The bigger takeaway is even more important. Google, Microsoft, and Anthropic all released capabilities this week that make autonomous, business-ready AI workflows far more practical than they were even a few months ago. For leaders, that means the window to experiment is still open—but it may not stay open for long.

In this session, you'll discover:

• Why AI is increasingly targeting work itself rather than the software layer around it
• The difference between intelligence work and judgment work, and why that matters for business leaders
• What Anthropic’s Claude usage data reveals about the gap between AI capability and actual adoption
• Why friction—not technical limitations—is slowing AI transformation inside companies
• How Google’s new Workspace CLI expands agent access across Gmail, Drive, Docs, Sheets, and more
• What Microsoft Copilot Cowork could mean for enterprise automation inside Microsoft 365
• Why AI review systems will become essential as AI-generated output scales across functions
• How autonomous agent loops could reshape software, marketing, sales, customer service, and product development
• What recent layoffs at Meta and Atlassian suggest about the future of knowledge work
• The legal battles emerging around AI, from copyright to legal advice to data privacy

About Leveraging AI

• The Ultimate AI Course for Business People: https://multiplai.ai/ai-course/
• YouTube Full Episodes: https://www.youtube.com/@Multiplai_AI/
• Connect with Isar Meitis: https://www.linkedin.com/in/isarmeitis/ 
• Join our Live Sessions, AI Hangouts and newsletter: https://services.multiplai.ai/events

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

Transcript

Speaker 0:00

hello and welcome to a Weekend News episode of the Leveraging AI Podcast, a podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. This is Isar Metis, your host, and we have a fascinating news episode for you today. I think they're all pretty interesting, but this one is really, really interesting. We're gonna focus. On two different main deep dive topics. One of them is going to be the impact of AI on the future of work from two different sources. A really interesting article or a blog post from Sequoia. And the other one is an analysis that was done by Anthropic on how people actually use Claude versus how they could be using Claude. And on the second deep dive topic, we're going to talk about several different launches that happened this week that when you connect the dots between them, show how profound the new world we live in is versus the way it was just a couple of months ago. And so we're gonna talk about what these releases are and why do I believe they are so dramatic in their impact on how we work versus how we should work or can work right now. And then we have lots of rapid fire items. Many of them are really interesting, maybe legal cases and legal implications this week. So we're gonna have a lot of those, which is a little surprising. Some big updates on job losses, so we have a lot to cover, so let's get started. The first story, as I mentioned, comes from a blog post on Sequoia's website that is called Services the New Software by Julian Beck, and it opens with a sentence that's saying the next$1 trillion company will be a software company masquerading as a service firm. The premise behind this is saying that for every dollar a company spends on software, they're spending$6 on services. Meaning the total addressable market for replacing work versus replacing software is significantly bigger. Now he goes into analyzing this in several different ways, and they're all very interesting, and I'm gonna give you a few examples of what he's talking about. But the first thing that he's looking at is how deep the penetration of AI is right now into different professions. What he's showing, and we'll see similar information coming from Anthropic themselves. In the next topic that we're going to dive into is that currently in software engineering? 49.7% of work is done with ai. But then the second one is back office automation, which is at 9.1%. So single digits and all the other aspects, including marketing and copywriting, sales and CRM, finance and accounting, data analysis and business intelligence, academic research, cybersecurity, et cetera, are all in single digits, and actually most of them in very low single digits. So it goes from 9.1 to 7.1 to 4.4, and then below that as you keep on going very, very quickly with legal being below 1% of usage. So the current level of usage that he's describing is very, very low. But then he creates a framework through which he defines as the opportunity map for AI in the future. And he has two vectors through which he's measuring it versus it's outsource or insourced. And versus it is judgment work or intelligence work. And I'm gonna read a short segment from that article for you to discuss the intelligence versus judgment. So he says, and I'm quoting, writing code is mostly intelligence, knowing what to build. Next is judgment. Translating a spec into code testing. Debugging the rules are complex, but they are rules. Judgment is different. It requires experience and taste. Instinct. Built on years of practice deciding which features to build next, whether to take on tech debt, when to ship before it's ready. What he's basically saying is that there's different kinds of work. Part of it is just intelligence, which is doing stuff, and the other is making decisions based on judgment on what is the right thing to do. Now I'm going back to the quote a year ago. Most cursor users treated AI as autocomplete. Today, more tasks are started by an agent than by humans. Software engineering accounts for over half of all AI tool usage across professions. Every other category is still in single digits. the reason is that software engineering is primarily int intelligence work, AI has crossed the threshold where it can do most of the intelligence work autonomously and leave the judgment to humans. Software engineering just got there first. It is coming for every single profession. So then he created four quadrants that describe where we are right now and where we are probably going. So on the bottom you have insourced. On the top you have outsourced. On the left you have judgment, and on the right you have intelligence. And obviously the top right corner where is what he calls autopilot territory is the outsource intelligence work. The stuff that AI can do right now and the stuff that is easier to let go because you don't have to fire your own employees, you just stop using outsourced work. And in that quarter, he has insurance, brokerage it, managed services, payroll compliance, claim, and adjusting accounting and audits, healthcare RevCycle, mortgage origination, K-Y-C-A-M-L, paralegal tax advisory, legal transactional real estate closing cost estimation, each and every one of them with the tens of billions or hundreds of billions of dollars of the size of that industry. Then if you go to what he calls the next wave is the section that is intelligence, but in-source, the stuff that you're doing in-house, which is obviously harder to let go of because these are employees of the company. That includes supply chain and procurement, pharmacy, back office, wealth management operations, medical admin, fund administration, and then on the top left corner with co-pilot territory, which is things that are outsourced, but require judgment management, consulting, graphics, UX design, executive search, PR and comms. And then the final frontier if you want the stuff that is in source and required judgment, which. Probably the furthest away has recruitment, advertising, freight brokerage, admin, assistant clinical trials, S-E-O-E-R-P, implementation, corporate training, market research, cybersecurity architecture, patent and IP work and travel management. I don't totally agree with all of that. I think some of it doesn't actually require judgment and is more intelligence. Uh, and some of it is actually outsourced, like travel management and patented and IP work. But I'm not gonna split hairs. I think overall the framework makes a lot of sense to me, and I'm gonna talk about more my thoughts on the general concept of judgment because it's very much aligns with what I've been saying, for the past few months, but we'll get to that later on. So the bottom line on what he's saying, he's saying that AI is going after the actual work itself and not the software that is doing the work that has been obvious for a while. Only he has a good framework to understand why things are happening before other things, again, outsource, versus insourced and judgment versus intelligence. But what he's also claiming is that AI systems accumulate proprietary domain knowledge because they are starting to do the intelligence work and hence they're going to develop judgment, which basically means that today's judgment work will become tomorrow. Intelligence work, meaning AI will learn how to do the judgment aspect of the work better than most humans because you will have access to a lot more past examples and understanding what good decisions or good judgment looks like on a much higher scale than more or less any human on the planet. Which basically means that eventually AI will be able to handle 100% of knowledge work without any human intervention better than humans. And he didn't mention exact timelines, but it's very clear that's not gonna take decades. We're talking about several years. Surprisingly, in the same week we got a new report from Anthropic that was trying to understand the gap between what Claude can do right now versus how it's actually being used based on people using Claude in conversations inside of the cloud platform. They were also trying to understand what is the impact on that, on real jobs in the real world right now. So let's start with the job aspect of it. What they're seeing is that job finding rates for workers ages 22 to 25 have been significantly impacted and have seen a 40% decline in the post Chachi PT era. However, they're saying that overall they're seeing very little displacement, specifically based on skills in which AI is good at. And all I can say right now for this is yet, I think we saw the early signs of it in 2025. I think we're gonna see a lot more of that in 2026, and we're gonna talk about that in a few minutes. But the other aspect of the research I find even more interesting and that shows why I think we're gonna see a lot more layoffs in 2026. The most interesting maybe finding that is the easiest to understand very, very quickly is this radar kind of chart that they have included in their findings. So it shows multiple different professions or if you want industries in the circle around. And then there are two shapes that shows you the percentage of capabilities of ai. One of the shapes is in blue that is showing the theoretical coverage of AI off the tasks in each and every one of these professions. And then in red, a much, much, much smaller shape in the middle of the actual covered usage of ai. So as an example, this chart is showing that the ability of AI to do tasks in management, business, finance, computer, and math, and architecture are above 80%. And for computer and math, business and finance and management, it's above 90%. And yet, when you look at how much it's actually being used, it's looking in computer and math at less than 40% for business and finance at around 30% and everything else significantly smaller. The other topic, which has a very high percentage, which is office and admin, Which is over 90% has a usage again, of lower than 40%. So even in the aspects that AI can theoretically do right now, extremely well as far as being able to successfully complete all or most tasks in these topics, the adoption in the industry is actually very, very low. Now another aspect that they looked at, which I found very interesting, is the gap between people. Positions which are highly exposed to AI versus the ones who are in positions who are less exposed to ai, and the gaps are incredible. The highly exposed workers are 16% points more likely to be female, earn 47% more than the average, and are 3.8 times more likely to hold graduate degrees compared to the unexposed workers, which basically telling us something that I've been saying for a long time now, that there is a big concentration of exposure in higher paid professions. Meaning the people with graduate degrees who are making significant amounts of money have a higher chances of losing their job or having their job changed completely because of AI adoption. Now they did acknowledge that their research is obviously not perfect and that it has failures similar to what they've seen in the past. An example that they're giving is that there were research done about a decade ago about the offshore ability of different roles in the US and it identified 25% of US jobs that could be displaced, and yet most of these jobs has maintained healthy growth in the US through that decade. I'm not sure that this is the same case right now, but it is definitely a valid point because in both cases it is replacing us employees with a cheaper version, cheaper option of doing this. I think there's a very big difference between using AI versus offshore. I had offshore employees most of my career, both in Eastern Europe and Southeast Asia. I have, People in Southeast Asia right now that are working for me in both my companies. And I can tell you that working with multiple AI agents the way I'm working right now give me a lot more control and clarity on what's happening at every given moment versus working with offshore teams where you get exposure to what they're doing and how they're doing on specific points. I'm not saying it's bad. I love working with my offshore teams, on both companies. I have incredible people who are doing amazing work. I'm just saying that my ability to know exactly what's happening and redirect what's happening in real time at any given moment is a lot easier with my agents than with an offshore team. the other aspect is obviously the spread of cost is significantly bigger. Meaning if you can save 20, 30, 50% by offshoring, you can save 95 to 99% by using agents. The temptation to go down that path versus the other path is significantly bigger. So what does that put us? To summarize this first segment of the show, I think we are in a situation where it is very, very clear that AI is going after work and not going after software. It currently has the ability to replace a huge portion of the work done in our economy right now, and the only thing that is slowing it down is friction. Friction. That is created from implementation, from hr, from legal reasons, from regulations, from just adoption timeframe, skilling and reskilling. It limitations, data security, all these kind of things that is slowing the implementation down, but the technology from the capability is already there and it's just amount of time until despite the friction, this gets implemented on a much larger scale, which will drive significant layoffs and significant changes in the way we work. And now to some interesting releases that happened this week and how they pour a lot of rocket fuel into the fire we just talked about. So the first topic I want to talk about that I'm personally really excited about, and to be fair, I'm excited about a lot of these new releases because they all have profound implications of how we work together with ai. But the first one is Google just released the latest version of their Google Workspace, CLI, which is a command line tool that giving both human developers or just regular humans and AI agents direct access to everything in workspace or Gmail, drive, calendar docs, sheets, chat, basically everything that is available on the workspace API. So let's define what that is for a second. For those of us who are less technical, there are several different ways to communicate with a piece of software from external, the. Historical most common one was the API and API is basically a language that allows other software to talk to another software. It has a schema, basically the definition of the language and the definition of the different functions that the API can do. You can read disinformation, you can write that information, you can sort this, you can do that. Basically all the different functions that exist in a software that is exposed to an external tool. It requires knowing how to write code or at least knowing how to ask a AI tool to write code. But it's still clunky and complicated and not easy to use for somebody who's not a developer and for somebody who doesn't wanna spend a lot of time. Then we have C-L-I-C-L-I is basically command line interface. It is. Ridiculous to me that in the year of 2026, it became the coolest thing because that has been dos, right? It's a command line interface that existed since we started communicating with computers, but it's making a very big comeback. The reason it's making a very big comeback is because it is allowing you with a single line to create an action that activates one aspect of the API one command that does something. The third aspect is that became a huge hit in the past year and some change is. MCP servers. And what MCP basically is it's a wrapper around the API, the biggest advantage of MCP is that with one line of code or a few lines of code that you can copy and paste off the internet, you can connect to the API of a different software. It has several disadvantage. The two biggest disadvantage, one, it's a layer of overhead above the regular API. So there's another step between you and the API itself and two is that it takes a big chunk of the context window of your AI tool, whether it's an agent or just a regular chat. So yes, it allows any person in the world to connect to many, many different software, which is magical, but it has a price. The CLI provides the best value for effort, right? The best ROI. You don't need to know anything, and yet you can call a lot of functions within the third party software. In this particular case, any aspect of the Google workspace, which is extremely valuable from a business perspective. The cool thing is to get this whole thing going. So let's say you are completely non-technical and you have no clue what you're doing. All you have to do is go into cloud code as an example, and type NPM install dash G, Google Workspace forward slash cli. That sounds really scary, but that's all it is. It's one thing, and from that moment on, you have access to create. To find information, change information, and create documents or anything you want inside the workspace environment. This allows anyone, literally any person to do incredible things inside Google Workspace. In addition, to make it even more attractive, Google shipped this with more than 40 prebuilt agent scales that are covering a huge spectrum of activities you would want to do inside of the Google Workspace, so you don't even have to figure that out. It just comes with it out of the box. This includes things like sending emails and managing calendar appointments, and filtering chat messages and filtering email, two different categories and so on. All of that comes out of the box with one line of installation, and then you can grow from there and do whatever you want. Now to make it even cooler, the way they've built this particular CLI tool is, instead of hard coding, it, it's actually built on top of a dynamically runtime that is reading Google's discovery service. Meaning as Google adds more functionality to their API, that functionality will immediately become available to your agent. So anything that an API can do in the Google Workspace environment, you can do from your agents, whether it's Cloud, cloud, cowork, or any other tool that you're using in order to build these kind of solutions. Now they've also done something that I find very smart is that they integrated it with Google's Cloud Model Armor, which scans the workspace API responses for prompt injections before they actually get to your agent. Which means it dramatically reduces the chances of malicious agents taking action and giving you answers that then can hijack how your agent actually works. So from a security perspective, it is much more secure than it was without this feature. Now, in addition to that, Google have released a whole new version of their Gemini features inside of Google Workspace. So Gemini, for Gmail, Gemini for. Google Sheets, Gemini for Google Drive, et cetera. All of these became significantly more capable than they were just a week ago with much better integration between all the different tools. So they're aware of your context across the entire workspace, and they have a lot more functionality and things they can perform, and for somebody like me who runs all my businesses through the Google Workspace environment, it is a huge benefit. I use those tools every single day, and having them be more capable and provide more capabilities inside the Google Workspace environment and from outside for agents is extremely valuable to me and to anybody who learns how to use them. Now to tell you why this is Important and impactful is that Google Workspace surpassed 3 billion monthly active users globally in the beginning of 2026, 11 million paying business customers and over 8 million paid Gemini Enterprise seats deployed across over 2,800 companies. This is the size of the install base, which is massive, and so every one of these changes that Google is deploying is impacting a very big chunk of the workforce out of the box. They don't have to get new clients in order to have AI create a very significant impact. All it requires is for companies to understand how to use the tools that they already have access to. Now in the past few weeks or maybe even couple of months, we've been talking a lot about Anthropic and Claude and talking a lot about open ai and especially with the whole situation with the government and the Department of War and the relationships between them and stuff like that. And we talked about 5.4 and Opus 4.6, but we haven't talked about a lot about Google, but they have been consistently releasing highly valuable solutions and capabilities into the Google environment that drives immense value. If you're in the Google universe and if you are, all you need to do is learn how to use this and you can drive incredible efficiencies. Right now with AI across everything, Google, but from my perspective, the biggest event that happened this week is actually Microsoft launched of copilot cowork. So what is Copilot Cowork? Copilot Cowork is Claude Cowork, maybe the most powerful, most user friendly AI tool in the market right now built into the Microsoft 365 environment. So it is taking copilot from a tool that can provide you answers to a tool that can work towards a goal autonomously by defining a plan and then executing it while having access to everything Microsoft Copilot has access to. So, a few really important things about this. Now, how does this exactly work? Microsoft has developed for copilot what they call work iq, which is a layer of data that makes all the relevant knowledge in your Microsoft ecosystem available to Microsoft copilot, cowork basically rides on top of that. So it can do everything that Claude Cowork can do, which you give it a task, you define the goal and it knows how to plan, find information, research it, synthesize the information, refine the plan, and execute all the different steps. But it also now has access to your entire knowledge that exists in your Microsoft environment and can act on each and every one of the Microsoft tools, including emails, meetings, files, et cetera, et cetera. Everything that exists in your Microsoft environment, it can interact with and also operate just as if it was a human operator. That makes it significantly more powerful than even Claude Cowork on its own. Or if you want, if we now connect the dots, what I can now do in Claude Cowork together with the Google CLI capability, so combining my access that I now have, that I just got from Google to allow cloud cowork to work in everything in the Google environment you can do inside the Microsoft environment without ever leaving it and without having to use third party tool like Claude Cowork. The biggest benefit of that is obviously the enterprise grade security and control that comes with working inside of the Microsoft environment. So why do I believe this is the most significant use of the year in the AI space so far? I told you several times in the past few weeks that I now spend 95% of my time where I'm not in meetings or doing training or stuff like that in the Claude Universe between Claude Cowork and Claude Code. Several of my clients are in the Microsoft universe and they had very serious issues with incorporating these things because they are limited from an IT security perspective to the Microsoft environment. And hence everything I wanted to show them or drive them to do, they were rejecting because of that reason that just went away merging the capabilities of the. Underlying infrastructure of Microsoft together with the incredible capabilities of Cloud cowork allows any organization that is in the Microsoft environment to automate more or less any knowledge work that exists in their organization right now. And as I mentioned, it's even better than Cloud Cowork itself because it has seamless access to the entire knowledge and ecosystem of Microsoft 365. So I'll repeat what that means. It means that every person with the right training in any organization in the Microsoft ecosystem can create highly sophisticated, multi-agent orchestration solutions that can automate more or less any knowledge work in your company right now. Now I've built a specific dedicated training plan for individuals and companies to train exactly how to do this. That includes best practices, infrastructure, examples, step-by-step learning from knowing nothing about what an agent is to being able to develop multi-agent solutions. And I already have several different companies that book me to deliver that training for them. I'm planning to open a course for that for the general public as well. But if you are a leader in a company, and you are in the Microsoft universe or in the Google universe because there's now solutions for both, and you want your company to be able to run at the speed of light compared to what you're running right now, please reach out to me on either LinkedIn or my email. There's a link to book time with me on my calendar, straight from the show notes of this podcast. This is a very significant moment in history of AI in businesses. And if you don't wanna stay behind, find a way to get your people trained. It doesn't have to be me. You can find other people to do this, but if you are listening to this podcast and you have any influence on your own future and or on your company's future, don't miss this opportunity because the gap between the people who know how to do this and those who don't is gonna open really, really fast. Now to add more to the things that were released this week that have significant impact, anthropic, just launch Claude Code Review inside of Claude Code. What it basically means is it means that you can launch a code review for every pull request for every new piece of code or updated piece of code that you are creating. and Claude Code Review will review it for you trying to find bugs, issues, security aspects, and so on, automatically, completely autonomously. Now, inside of Anthropic, they're saying that now code review appears in 54% of pool requests inside of Anthropics code, meaning more than 50% of new code that gets checked into that gets committed in the new code Inside of the Anthropic universe has changes made by code review. This is very significant. It's up from 16% before the implementation, meaning it's really, really good at finding bugs, and they're saying that it is identifying bugs in 84% of large pieces of code that has more than a thousand lines changed in them. The other thing is they're saying it's doing it with very low rate of false positive, so it's finding on average 7.5 issues per review with less than 1% of false positive, meaning it's very rarely identifying bugs that are not actually bugs. There's been mixed feelings about the code review feature when it was released. Many people are saying it's extremely valuable and powerful. Obviously, a lot of people inside philanthropic said that it's completely changed the way they work, but it is currently billed based on token usage, which means every review costs between 15 to$25. That means that if you're shipping a lot of code, this could become very expensive very quickly, and then you need to decide whether it's worth it or not. So let's describe for a second the problem that it's solving, and then let's generalize what the hell that means. The problems that it's solving is that the amount of code that gets generated right now is orders of magnitude more than the amount of code that was created before. AI created most of the code in the world. So the people who are reviewing the code were, by the way, in many cases, the bottleneck before this case. But now it is crazy, right? So if you had 10 developers and three people reviewing the code, it was more or less doable. Now you have 10 developers creating the amount of code that a hundred developers created before, and you still have the same three people to review the code. They're becoming a very significant bottleneck, which means, A, they slow the deployment down, and B, the chance that they will miss things because they got to run faster is very significant. And C, it is just not doable. There's just not enough people in the world to review the code that will be generated in the very near future, meaning the ability to review the code by AI is necessary. The entire cycle of development of software has to change from end-to-end, because from the design process, which I'm doing right now with ai, without a design team, without product managers, but with AI agents in these roles has changed completely. And every company that creates code or any company who wants to create code needs to learn new ways of doing that. So is 15 to$25 per review high or low, or too much, or not enough? I'm not sure. I'm sure companies will have to figure it out on their own, but this kind of solution is mandatory moving forward because otherwise code is just not going to be reviewed, which is obviously not a good idea. Why do we think this is significant? Because the same exact thing is coming for any kind of work. We can now create a hundred x more marketing content, what we ever could. Somebody needs to review this marketing content to make sure it is a valuable B, aligned with our core values. C aligned with our brand guidelines, D aligned with the goals of the campaign that we're running, et cetera, et cetera. Humans cannot do this because there's a huge amount of it that gets created at any given point. The same thing applies to more or less any outputs that a company can generate in the digital universe. Which means we need to learn how to build AI review systems to review the things that other AI systems are generating to make sure that what we're shipping, regardless of which aspect of the business it is, is actually what we want to ship, and it's not gonna create any issues with our technical issues, legal issues, customer service issues, brand issues, and so on. We will have to learn how to do this across every aspect of the business. But let's continue with additional interesting releases this week. Andre Cari, who is one of the co-founders in OpenAI. He was one of the leaders of the Tesla AI team, and he's a highly influential in the AI space. We talked about him many times. On this podcast. He has just released something interesting, which is a minimalist open source python library called Alto Search. It is a very short piece of code. It's, have only 630 lines, but what it does, it is a short loop that allows to train small AI models on a single GPU overnight. Now, the vast majority of us listening to this podcast, myself included, don't have any plans of training our own models. So why am I including this inside of a deep dive section of this podcast is because of what it means. So let me explain to you a little bit what this thing actually does. What this thing does is it takes an existing set of parameters of a small, large language model, and then it uses two files. One is a markdown file from a human, the person who wants to run it, who defines the goals and the research instructions and so on. And another file that is controlled by the ai which is a library that the AI itself writes as it's learning new things. Every run runs for five minutes and then there's a clear defined criteria on how to score the output. So then it goes and compares the output to the previous version that existed five minutes before that to see if it's a better version, which means it now becomes the baseline, or it is not as good as the previous one, and then it goes down the trash. But because the process itself is taking notes and learning from one cycle to the other, it is actually getting better and better from one cycle to the next. And it can develop and optimize a large language model over a small language model overnight. So if you think about it, because every experiment runs for five minutes. This means it can runs about 12 experiments per hour, which is about a hundred experiments in a single night, which allows it to learn very quickly on its own. But this is just step one. The second step that Andre is planning and some people have already implemented, like Varun Martha, who's the CEO of Hyperspace, is connecting multiple of these agents to a single shared knowledge base where they can learn from each other's experiments and not just each agent learning from its own experience. So between March 8th and March 9th, he had 35 autonomous agents on the hyperspace network run 333 fully unsupervised experiments. Now, because they are sharing a single knowledge base of learning from one another,. When one agent discovered something, it told the other agents, which then knew how to do this thing without having to get to that in their own experimentation. This happened in several key findings through the night. So one agent's success became the knowledge for every other agent in that cluster, basically replicating how human research communities work. Only when research communities work, it takes a very long time for them to share information because they have to share research papers that gets published not every single minute, and they're, they have to be conferences for them to discuss this and so on. You're talking about unlimited number of these that are running 24 7 that are sharing information with one another in near real time. so let me give you the final quote that I wanna share from Andre and then tell you what I think about everything we talked about so far in the show. And then we're gonna go into rapid fire. Andre said, the goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. The only word that is unnecessary in this sentence, if you wanna generalize it, is the word research. So if you generalize this, it will say, the goal is to engineer your agents to make the fastest progress indefinitely and without any of your involvement, not just in research. So why do I believe that this is a. Highly critical aspect and why this puts together and connects a lot of the dots that we talked about today and in the past few weeks. If you remember a few weeks ago we talked about Ralph Loop. So what is Ralph Loop? Ralph Loop is a very simple tool, actually a few simple lines of markdown file that allows to do the following. Look at a list of tasks for development in software. Pick the one that is the most relevant to develop. Next, send it to a set of agents that will write the code, send it to a different set of agents that will review the code, send it back to the agents to fix the code in order to solve the bugs. Complete all the bugs that are there in several different cycles. Once the bugs are solved, commit that code and then the loop starts over again by looking for the next most relevant tasks and going and developing it. This is a very similar approach to what Andre Carpathy just released. It's a loop of agents with a clear score in the end to identify whether what you did is the right thing that needs to actually happen. But once you scale that with multiple of these loops in parallel, that share a knowledge base with each other and they can run simultaneously while learning from the mistake and successes of all the other agents. This is now multiplies the capability of these loops almost indefinitely. And you can do this for any field of business. So here are a few examples in the marketing space. Think about optimizing any marketing copy, whether it's newsletters, landing pages, brochures, website, et cetera. You test it, you run it, you evaluate it. If it's better, you let it run. If it's not better, you kill it, but you can run as many experiments. As you want at any given time in sales. Think about writing proposals, evaluating the proposals based on different set of metrics, and then rewriting it until it is fully optimized for the specific target audience, or based on the specific RFP that you are using. In order to write the proposal in customer service, think about considering multiple options for the best way to handle a specific situation, and then for the best way to communicate it with a client in product development. Think about considering multiple options for the next steps. The next features, the next cycle, the next version, and optimizing for different outcomes in drug development. Think about testing a huge variety of options before actually deciding on the ones that you actually want to pursue and so on and so forth. For almost anything in any industry, having an army of agent that run multiple tests while scoring them and evaluating them and picking the winners while learning from the mistakes can accelerate the path. Two success by orders of magnitude compared to how we are working right now. Literally, anything you can objectively measure and score, you can apply this model to. Now, since we already have open source solutions that built the scaffolding, the infrastructure for this, and I'm sure there's gonna be even better of them coming soon. Any company can apply this to any aspect that they want. In addition, I'm pretty sure that very, very quickly, this will become a easy to use infrastructure where it won't require you to know how to go to GitHub and use a repo, which is a lot easier than most people think, but very soon there is going to be a user friendly version of that tailored into every aspect of our work as part of the tools you are using today. And that means that any person in any company doing any knowledge work will be able to apply these kind of loop operations that are self-optimizing across everything in the company. I'm already doing a lot of this stuff in the agents that I'm developing, and in the course that I'm working on is gonna be one of the topics, how to have a unified knowledge sharing across agents across every aspects of the business so they can learn much faster and not repeat mistakes that other agents have made, or in the development of other agents. Now let's connect all these dots. The convergence of all these different capabilities such as the CLI, access to your entire Google workspace, such as co-work inside of your Microsoft environment, such as the ability to run, a huge number of experiments that self-improve to deliver specific business outcomes are complete game changers compared to what we knew just a couple of months ago. Now we are not even starting to understand what this actually means for the way business is run today, but one thing is very, very obvious to me, and if you're listening, I assume it's obvious to you as well. Those who will implement this now will gain a huge competitive benefit, at least in the short to medium future until everybody else figures it out. And again, if you wanna learn how to do this, please reach out to me, either as an individual or as a company. You can do this, as I mentioned on LinkedIn or through the links to my calendar or my email on the show notes of this podcast. Now let's dive into the rapid fire. We have a lot to talk about, so let's start with. The impact of all of this and other aspects on workforce. Meta is planning to lay off up to 20% of its workforce. That is close to 16,000 employees that are going to be let go out of the 79,000 employees that Meta has Worldwide. Meta is planning this in order to do two things. A, to compensate for the huge amount of investment they have in AI in the future. They have committed over$600 billion to build data centers by 2028. They just spent close to$2 billion on Chinese startup Manus. They are offering million dollar compensation packages to hire new people to their super intelligence team. So they have huge expenses on the AI aspect. So far without a lot of new income on the other side, and so they have to find ways to compensate for that. The other reason, if you remember, mark Zuckerberg stated in January, that he's seeing projects that used to require big teams now be accomplished by a single, very talented person. So the combination of these two things leads to this really sad and yet obvious outcome where they gonna lay off 16,000 people sometime this year. There's no any formal thing, but it has already been communicated to the leadership team and through the leadership team to the second tier leaders inside of Meta that this is coming. To stay for a second on the meta topic before I continue to additional layoffs. In this particular case, in Atlassian meta announced that they have acquired Malt Book. So if you remember, we talked about Malt Book just a few weeks ago. Malt Book is a social network for agents built by Open Cloud. That was for a very short time, was called Malt Bot. so that's why the name Malt book. So it's the combination of Malt bots together with Facebook to create this platform. So Meta just acquired the platform and if you want, most importantly, they acquired the two people behind it. Are Match Lint and Ben Par? The specific announcements from Meta Says, and I'm quoting the Malt Book team, joining MSL opens up new ways for AI agents to work for people and businesses. Their approach to connecting agents through an always on directory is novel step in a rapidly developing space, and we look forward to working together to bring innovation, secure agent experiences to everyone. So that's the announcements from Meta's Spokesperson. If you remember just a few weeks ago, OpenAI has hired, or if you want, purchased Open Claw and its creator, which is a thing that started the whole craze that then drove these two people to create notebook. So what the hell is going on here? And why would Meta and OpenAI hire people who developed a piece of technology that they can replicate in about 10 minutes? So on the meta aspect. Something that I think plays a big role is that Notebook is maybe the only quote unquote social network, new idea that came online and became successful in probably at least 10 years. So that by itself is worth acquiring, just so that nobody else captures that, and then they might face some kind of a competition in the future. But I believe there's something much bigger than that that ties to something I shared with you several times in the past few weeks. But I will share with you again because I think it is very important that you understand this, and or if you miss this in previous week. The world of knowledge. Work has shifted from a world in which the limiting factor for growth of a company is resources, money, compute people, et cetera, to a world in which you need three things in order to drive 10 X scale to any knowledge business, great ideas, good judgment, and very well defined requirements. If you have these three things, you can literally build anything you want because the agentic tools allow you to build these kind of solutions. But for these solutions to be successful, you need to have the right idea. You need to have the right judgment in order to A, prioritize ideas, and B, then guide your original idea as it's being developed to the right direction to provide value and become successful. And you need very, very clear requirements in order to be able to develop this in a way that will actually work and will be aligned with the vision. The people who are just hired by AI and Meta obviously have all three at the highest levels. They had a great idea, they had solid judgment, and they were able to define clear requirements and create the product or the service that they wanted to create. These are exactly the kind of people that every company needs to hire right now. Every company that will be able to get to put their hands on many of these kind of people that can have at least one and hopefully two or three of these capabilities. So the ability to come up with great ideas, great judgment, and the ability to create detailed requirements to enable the execution to happen through agentic processes. It'll be rocket fuel for the future of these companies. I am 100% focused on this myself. I already see incredible results for the things from the things I've developed in just the past couple of months following this same exact concept. And if you are a leader in a company, this is one of the things you have to start thinking about in the kind of people you want in your company. So back to the impact on workforce Atlassian, which is an Australian company that is. One of the key players in the software collaboration tools with tools like Jira that more or less every software company in the world uses in order to define requirements and track bugs and so on, is just announced that they're going to let go 10% of their workforce, which is going to be about 1600 employees that are gonna let go with an aim to, and I'm quoting, reshape its skill mix and in order to offset their huge investments in AI enterprise solutions. So very similar to what we hear from meta. It is a double digit percentage of the employees that are gonna get let go in order to support these two things. It is very obvious to me that this is gonna keep on happening and it is currently happening in the software space more because this was the first topic in which the AI platforms that invested but connected to everything we talked about in this episode. This is coming for every other knowledge work. Quick update on the topic that we discussed in detail last week, which is the clash between philanthropic and the government, and specifically the Department of War A. It there was an announcement that said that there is no way this will be reversed and come back to a successful outcome of new negotiations. But on March 9th, philanthropic has filed two federal lawsuits against the US government alleging that the supply chain risk delegation and the President Trump's February 27 ordered to federal agencies to seize using its technology constitutes illegal retaliation for its protection of speech and violates constitutional and stationary law. How is this going to evolve, how it will impact the actual detachment of Anthropics tools from the government? I'm not exactly sure, but I will definitely report as things evolve. Staying on the legal world, and I told you a lot of interesting things happen in the legal space. This week, a lawsuit has been filed against OpenAI by Nipon Life Insurance Company of America, alleging that Chachi's chatbot engaging unauthorized practice of law by providing legal advice to an individual who subsequently used it to file numerous motions and a new lawsuit after a settled disability claim. So the person, to Cie del to who is a former disability claimant. Allegedly was unhappy with the settlement that her lawyers got her in a lawsuit and she has uploaded all the previous data of her previous trial from her lawyer into Chachi pt that validated her concern about being gaslighted and encouraged her to fire the lawyer and then assisted her in drafting dozens of motions and a new lawsuit despite the fact that the original case has been settled with prejudice. Now to be fair, opens AI's user policy from October 29th, 2025 explicitly prohibit using its services for tailored legal or medical advice without a licensed professional involvement. That being said, the lawsuit is still there and PON claims that this is too little and too late, and OpenAI is saying that this is all bullshit and they don't understand why they're being sued. I assume the legal system will figure out a way to address that. That being said, this situation where individuals are using these tool for either legal work and or medical advice is growing every single day, and there is now legal work to try and block that. As an example, there's a bill advancing in New York, which aims to explicitly prohibit AI chatbots from dispensing substantive. From creating legal advice, Annie's planning to create private right action for users harmed by such conduct. But combine that with the fact that the federal government is going to try to block every state kind of motion moving forward. I'm not sure how helpful this is going to be. It is definitely a problematic situation right now as somebody that has been using Claude and Cji, PT and Gemini excessively to review and write legal documents. I will be really sad if this is being blocked because it's saving me a hell of a lot of money. If there's lawyers listening, I'm sure they think I'm completely crazy. But as somebody that has reviewed legal documents for three decades right now generated by lawyers, I do not see a big difference between the great advice that I'm seeing from these tools compared to what I paid for a lot of money by lawyers. I still use my law firm for final review of big things, but not for everything, and definitely not from the initial state. Staying on the legal field, Elon Musk's lawsuit against OpenAI is entering a new phase where on Friday last week, federal judge has decided what evidence and expert testimony can proceed to trial in April. Now, to put things in perspective, this case is going to decide whether the valuation of Musk's early donations into OpenAI in the very early stages is currently worth up to$109 billion, which is what Elon is claiming, and that is going to be reviewed and decided in court starting next month. But despite all the backlash on AI in the legal world, Swedish AI legal platform leg, Agora has just raised$550 million in a Series D with a five point$55 billion valuation tripling its value from just a few months ago in their series C, another company, LA Gore's largest competitor, Harvey, has the same kind of situation. This is a US based company that is backed by A 16 Z already is already valued at 8 billion and reportedly seeking 11 billion in its next round. Both platforms supply a different kind of service. They actually provide AI solutions to law firms and legal companies to execute more effectively, or as the CEO of Laora said. It is amazing that everybody can have their own pocket lawyer in Claude, but we are not solving for the same use case. Staying on the legal world, meta is now facing a major class action lawsuit after an investigation by a Swedish newspaper revealed that workers in a Kenya based AI subcontractor company are reviewing footage from the mes, RayBan sunglasses, including some highly intimate personal situations that were recorded by these glasses. The lawsuit doesn't actually go after the fact that this is being done, but after the fact that the campaign for these glasses focused on designed for privacy, and that is obviously false advertising and not the case when your intimate moments, including nudity and sexual scenes are being reviewed by humans without your consent. To make this even worse, you have no means as a user of these glasses to block your content from being reviewed by humans. Both the visual aspect of it as well as the voice communications both are potentially being reviewed by humans, and in both cases you have no way of blocking that. Now, based on the fact that over 7 million pairs of these glasses were sold in 2025 alone, the potential aspect of this class action could be enormous. I said multiple times before that there are huge issues with these smart glasses from multiple aspects of privacy. This is just one aspect of your own personal privacy, but the fact that a person that's wearing them is basically recording everything around them without the consent of the people around them in any situation is very problematic. And I don't know if it's breaking any rules right now, but I think it should break several rules that should come up to protect specific scenarios from being recorded without your consent staying on legal topics. The Supreme Court just denied an appeal by a person named Fowler, who for the past few years have been fighting the windmills of the copyright process in the US to get a copyright protection for a work that he has done with ai. The story actually started in 2018 when Fowler had listed. AI as the sole author of his copyright application for a work that is called a Recent Entrance to Paradise, that was rejected by the copyright office. He then took it through the judicial system in its different tiers, all the way up to the DC Circuit Court of Appeals, who kept the copyright office statement in place. And again, now he was rejected an appeal to the Supreme Court, which leaves the statement as is the DC Court of Appeals affirmed the statement that says, and I quote, the court has said that calling human authorship a bedrock requirement of copyright, basically saying that AI cannot be at least the sole author of something that gets copyright protection. Now, the problem is how much human involvement is required for corporate protection remains unanswered. In a related case, that is Allen v. Perel Motor is currently in the US District Court of Colorado, where the artist Jason Allen, is challenging the corporate office refusal to register a work. He refined using over 600 AI prompts, arguing that his significant creative inputs distinguishes his claim. From the other quote where Thaer just gave one prompt and he's trying to copyright that. I tend to agree. I think that a significant input into a prompt or a set of prompts or a entire process of working with AI should receive copyright protection for two different reasons. A, you are bringing your own capabilities, your own creativity, your own mindset, your own judgment into the output, and B, if that is going to be disqualified, then copyright law is over. Because I think moving forward, most of human creation is going to be assisted by ai, and that means we don't have any copyright for more or less anything in the future. Switching gears to other topics. Repli just raised$400 million in a Series D and have unveiled agent four. Why is that interesting? Beyond the amount of money that they're raising? That now becomes common because we're used to seeing these crazy numbers that were science fiction just a few years ago. they're claiming that Agent Four, which is their fourth generation of agents, is now capable of vibe coding an entire company, an entire software company from scratch on its own. So not just a simple application, but think about Salesforce or something like that, that being fully created fully autonomously by these set of agents. Now to make this even more extreme rep is stating that Agent four is 10 times faster than the previous version Agent three that they released just a few months ago. So not only that, you can create very sophisticated software. You can do this faster and cheaper than before. A big new feature of that is called Canvas, which is a digital scratch pad that allows you to see the different elements of the project and move them around and annotate cards working collaboratively between human developers and the AI agents in order to have more control and more understanding on how your software project is developing. Now speaking on the software development space, interesting new announcement. XI. Elon Musk's Company just recruited Andrew Milledge and Jason Ginsburg, two senior leaders from the AI Coding Startup Cursor in order to enhance the coding capabilities of XI. This has been a major focus for Elon, who has admitted publicly that their platform is not at the frontier right now, but he said it's a gap they can close by mid 2026, which is just around the corner, and I'm sure that hiring these two people from Cursor is a push in that direction. Another interesting announcement this week that has to do with the ability to do things faster and cheaper is Amazon AWS just announced a significant collaboration with ai, chief Startups, Cerebra Systems, aiming to deliver their capability through the AWS Cloud solutions. Those of you who don't know Cerebra, they're the largest competitor of Grok with A-Q-G-R-O-Q, both these companies are developing. A different kind of chips than GPUs that specialize in inference. Inference is when the AI is being used to generate its outputs. So it is not good or not optimized for training models like the GPUs are, but they are much better, faster, and more efficient than GPUs in the delivery of ai, which is becoming a bigger and bigger portion of the AI compute that needs to be done because more and more people and companies are actually using ai. So in the first year or two of AI training the models to most of the compute, that is shifting very, very fast to usage. And these companies have very capable solutions. if you remember Nvidia, just Acqui hired the leadership of Grok, and now we have announcement from Amazon that they are going to be partnering and delivering the cereus platform to AWS users in combination with their own in-house chips. That's it for today. All that I can request is that you think deeply about what we talked about in the beginning of this episode. It will change. Everything that we know in knowledge work, and the capability exists right now. This is not a future capability. You can start doing these things this minute or tomorrow in your company. With the right training for the relevant employees. We will be back on Tuesday with a new how to episode that will show you how to implement a specific AI use case. If you are enjoying this podcast, if it gives you valuable knowledge, I would really appreciate it if you spend. Two minutes in clicking on the podcast platform that you're on right now, giving us a five star review or whatever you think I deserve, and then share the podcast with other people. There's a share button on your platform. It will take you two minutes to enter a list of several different people that you know that can benefit from this. They will be grateful. I will be grateful and you will feel good about helping people know more about ai. That is it for this weekend. Have a great rest of your weekend and I'll see you again on Tuesday.