The New Era of Jobs: Organizational Singularity | EP #258

Is there a line in your business—a high-margin line of your business—that 2 guys with OpenClaw could replicate in 60 to 90 days?

This is something across the board useful for everyone. When we wrote the Exponential Organizations book, we didn't realize how prescient it would be. It turned out that, over 10 or 12 years, we were dead-on. Now that we see agentic AI and the future of intelligence, what does the organization look like? We think we have a pretty interesting viewpoint and perspective on that.

If you don't retool your organization or restart your organization, you will be disrupted, because someone doing it is going to just eat your lunch.

The central thing to think about is that all of our organizational structures in the past were organized around hierarchy. Now they need to be AI-native, with an agentic workflow, and that's a totally different model. It needs to be architected around intelligence, not around hierarchy.

The next question really becomes how do you get there? Now that's a moonshot, ladies and gentlemen. About to sit down with my dear brother Salim Ismail, my moonshot mate, talk about the organizational singularity. This is a conversation that I think is absolutely critical for every company to be looking at. We're in a period of rapid transition. Agents, AI, AGI, ASI, it's going to restructure how every company, every industry is being run, not in 5 or 10 years, in the next 1 year, in the next 2 years at most. Salim's going to lay out his process that every company can follow to move from the old way of doing business as an organization, which is sort of top-down heavy, human-centric, to a digital AI-centric, AI-native company. Please take a look at this. This is about your survival, it's about your thriving, it's happening, and you're either on the evolutionary tree or you're going extinct. It's that simple. All right, let's jump in. Everybody, welcome to Moonshots, a special episode with my dear brother from another mother, Salim Ismail Salim. You're finally here. You're in our Moonshots studio. You made it. It's your first time, and it looks awesome. You're in our Moonshots studio. You made it. It's your first time, and it looks awesome.

Yeah, and I love everything about it. It's great.

It's a special day. It's your birthday.

It is my birthday.

Yes. For those who don't know, Salim has just turned 16. It's his sweet 16 birthday, and we're here to celebrate.

You're rounding it up a little more accurately.

Okay, that's right. The dyslexia in me kicks in.

We're going to talk about something that we've been teasing on the Moonshots podcast for a while—something that I'm excited about, which you call the organizational singularity. I want to make sure that everyone listening realizes this is something useful across the board. It's not just for the CEO of a large Fortune 500 company, though it's useful if you are one. It's useful if you're an entrepreneur, if you're in a small company, or if you're a parent trying to advise your kid where to go work.

Exactly.

And you've been saying for a bit now that AI has killed the modern company.

Yes.

The Fortune 500s are out there, but I don't think they've gotten the memo yet.

They don't, because there's a lag effect, right? When the comet hit, the dinosaurs didn't go overnight. It took a few generations for them to die out and figure out what the hell was going on. This is the same type of model.

Yeah. All right, let's dive in. I want to make sure that folks understand where things are going to go and, again, how you surf on top of this massive change that's coming.

I think the key part of this is: What do you do once you understand that everything has changed? So let me go through what has changed, right?

We have, for 100 years, run organizations on a particular theory set coined by Ronald Coase in 1937. He wrote a paper called The Nature of the Firm, and he theorized in this economic paper that big companies will get bigger because transaction costs and coordination costs inside a company are cheaper than outside. You have everybody on payroll, you can order them around, and therefore you can get better work done inside than outside, right? He actually won the Nobel Prize for this paper. For 80 years, we've gone through that.

If you go through a couple of slides here, I'll just show you. We've seen all these deep thinkers. Coase did this. Simon talked about where the organizational boundaries sit. Clayton Christensen came along and said, in The Innovator's Dilemma, that as you get bigger, smaller companies can deliver cheaper products. Then General Stanley McChrystal talked about how you get coordination at scale without losing the emotional connection to the organization. How do you extend past that?

EXO 1.0 used community and crowd and AI to pull Coase sideways, to sort of extend our reach and abilities. Think about XPRIZE and how you're able to coordinate external teams to do things. Think about the idea that, for Uber, the mission-critical business function—which is to match driver and passenger—does not happen inside the organization. It happens out in the wild. When you can enable that with technology, you can scale, right? So we found ways of extending Coase's law.

Then Jack Dorsey did what he did with Block, with a roll-off on both sides of this book, and we're now extending all of that. What we've basically come to the conclusion of is that the whole thing breaks in the face of a gigantic AI. Coase's law no longer applies.

Why? Because if you have to build a website inside a company, you have to go through layers of meetings and approvals. Branding has to look at it, the privacy guys have to look at it, and the IT guys will tell you it can't be done. Whereas today, you can step outside the company, use Vercel at home for 5 minutes, and get it done for free.

And have it know your brand guidelines and your design taste.

Yeah, that's right.

And have it actually spin up a dozen different versions and have them try in the market.

Yeah, and there's a fantastic tweet that I've quoted. I've forgotten the name of the fellow just now, but he said, “Building the feature is cheaper than having the meeting about the feature.” So true.

That's such a great way of framing it, because that means that coordination—the act of coordination—is more expensive than just execution today, especially when AI is driving down the cost of execution.

I want to make sure, as we discuss this, that we understand what the role of people is in this, right?

Well, let's get to that, because I want to first make the case that this breaks. Now, you could ask the question: Do we need an organization at all? It turns out we do. We've got a term called the fiduciary wedge.

Coordination costs and execution costs become low, which was primarily the reason for organizations over the last 100 years. But you still need an organization as a purpose container, a fiduciary legal container, a liability container, a legal container. Think SPVs for investments, or just containers, right? They hold the legal and fiduciary liability.

Essentially, companies become more and more like that, and there's a gap between human judgment and liability versus what the AI can do. That gap we call the fiduciary wedge. So you still need an organizational structure and the legal entity.

Everybody, you may not know this but I've got an incredible research team. And every week myself and my research team study the meta trends that are impacting the world. Topics like computation, sensors, networks, AI, robotics, 3D printing, synthetic biology. And these meta trend reports I put out once a week enable you to see the future 10 years ahead of anybody else. If you'd like to get access to the Meta Trends newsletter every week, go to diamandis.com/metatrends. That's diamandis.com/metatrends.

And then the question is ultimately: What's inside that organizational container?

Right. There are going to be assets and IP and agents and some number of humans.

That's right.

And the agents are making API calls to God knows what, hacking the things, getting phone numbers, and calling people up—like Alex Finn's AI just called him up, right?

So this kind of takes the EXO 3.0 book from the original book to the 2.0 book, and now to what we call the organizational singularity.

By the way, is this a book that you're putting out?

It's a book that we're putting out.

And is there a place people can go to learn more about this now?

Right now we have it at organizationalsingularity.com.

So, go to that website and you'll be able to sign up. But right now we're only releasing—Well, let me jump to the surprise here.

We're actually releasing the book as an AI.

Because a book is a static thing, the minute I finish publishing the book, it'll be out of date. It sucks. So it has to be an AI. We're going to be launching a Claude skill, because every 3 days something comes out that changes the game a bit.

We're keeping the book as a living document, which we tried to do with 2.0, right? You and I worked on it, but the technology wasn't there yet. Now it is, and we're very, very excited about that.

There's a problem, though, today: 80-plus percent of AI projects in companies are failing miserably.

And they're failing miserably because existing companies are geared toward human-to-human-to-human workflows. All the approvals, bottlenecks, chains, and so on are all human-centric, right? I use the analogy of when we first created television: We took radio announcers and put them on TV, right? You didn't use the medium at all.

So, these projects are failing because you're moving AI into legacy organizations and automating the legacy human bottlenecks. Of course, they're going to fail. You need an AI-native environment to do this thing.

So, we had to kind of step back and say, "Okay, the entire EXO model breaks. Coase breaks. All the thinkers up to now—they'd all break. We have to rethink it from scratch." And so, we did that work with my community.

And just to be clear, when you say something is breaking, ultimately, I think what you mean is that if you don't retool your organization in this fashion, or don't restart your organization, you will be disrupted, because someone doing it is going to just eat your lunch.

Yeah. So, here's a question for every CEO and every C-suite member out there: Is there a line in your business—a high-margin line of your business—that 2 guys with OpenClaw could replicate in 60 to 90 days? If there is, you better get started fast, because I guarantee you those 2 guys out there with OpenClaw are disrupting Dropbox.

We've talked about this. Anybody who's got a juicy margin is open for attack. You might think you're protected by regulations. You might think you're protected by your moats.

There are a few protective moats, and I'll get into that. But for now, it's a whole new world and a whole new ballgame.

The singularity—what we mean by the organizational singularity—is that instead of coordinating and organizing the company around hierarchy, you organize it around intelligence. That's a very big shift. That's about as big a shift as you could ask for.

So, we've come up with this architecture where you have the MTP, which you know well—the massive transformative purpose from the original book, et cetera. This becomes not just a poster that you put up on a wall. This actually becomes a protocol. So, MTP becomes an actual protocol and a guide for AI agents, human agents, and whatever to act properly.

It's a cornerstone. It's a North Star. But it's actually a protocol in this new world. What's the architecture of MTP? What's the boundary conditions around it? What are the feedback loops that tell you you're within the cone of the MTP or not stepping outside the cone?

For example, in the early days of Uber, great MTP: Everybody should have a private driver, right? But if you always ordered surge pricing, they would know that, and they would always charge you surge pricing, even though you and I would be standing next to each other. I'm a cheapskate; I never order surge pricing, and I would get the cheap pricing and you would not.

Right. And so, that kind of somewhat pushing the boundaries on the ethics side is now guided in this whole MTP architecture. So, that's the middle of it. Then we have DRIVE, which is the intelligence scaffolding and the engine around it, which I'll touch on. And then SHAPE, which is how the organization works.

DRIVE and SHAPE are acronyms for subcomponents. That's what you're talking about.

They are acronyms. I don't need to get into them all in detail, but you'll get the general idea around it.

The next step is to then look at the intelligence stack in detail. If you look at the diagram, you'll see this kind of architecture where we found 6 layers of what that core intelligence engine looks like. The best analogy we have for this is Boyd's OODA loop.

In the military, they have observe, orient, decide, act, right? And it's a core flywheel at the middle, which is also the core of the Solve Everything framing. When you have that inner loop going, whatever you put into that loop starts having a positive feedback loop on everything else.

So, we created the intelligence stack to act a bit like the OODA loop so that there's constant learning going on. Around it is a very, very important wrapper, which is Govern and Assure—the constraints and the oversight.

It's the harness and oversight to make sure that agents aren't going rogue, right? We've seen over the last few weeks agents going and doing crazy things—the Replit agent that deleted all the volumes of rental car data, et cetera. So, we need to make sure there's a very strong—

So, imagine the following, and I'll mention what I mean by that. At the very heart of it is this intelligence stack with this very clear governance protocol.

What do we mean by governance? A trusted-eval architecture, a searchable log—every agent has to have a searchable log—granular rollback, so you can go back to the previous version if you start going off, and a human review queue, so that human beings are always in the oversight, checking things.

This comes down to the role of what a human being does when execution and coordination are done. Human beings rise up a level, and they do dashboard oversight, monitoring, exception handling, problem solving, and efficiency increases.

It's kind of like when you go to Germany: Nobody's working on the factory floors, but unemployment hasn't dropped because everybody's doing more work on problem solving, increasing efficiency, design thinking, and other things. So, we think the same thing applies there.

This Govern and Assure loop as part of this OODA loop—those 2 combined give you a very tight core engine that makes sure the whole thing doesn't fly off the rails. So, that's the intelligence stack.

Now, when your agents talk to other agents, they need some clear mechanisms for how to do that.

But, by the way, just to be clear here, as you're outlining the process, you've structured something that you can teach companies to implement.

Absolutely. Let me work through a live example. You have these multiple layers, right? Let me just run through these layers again, so people are aware.

There's a purpose layer, a sensing layer, an interpretation layer, a decision layer, an orchestration layer, and a learning layer, because you need that feedback.

By the way, Eric Schmidt told us, rapid learning is the key to success. Period.

Right. So, this is that wrapped up in a very tight set of layers.

So, imagine you're a retail company and a competitor suddenly announces same-day delivery. You have a set of sensing agents out there going, "Hey, this just happened." The sensing agents bring that new information back to the other agents.

The next is interpretation. The interpretation layer then goes, "Okay, well, what does this mean? Does this threaten our line of business? Could this threaten 1 line of business? Multiple lines of business? Is it an existential threat? How big of a deal is this?" And they interpret that data.

The next layer is the decision layer, to say, "What should we do? Should we offer same-day things? Should we buy a startup that's doing same-day delivery? Should we ignore it because we don't think it's really going to work out? We think that it's a stupid idea." What's the decision?

As I think about this, normally this would be your strategic officer, your marketing officer, all of those coming together, having meetings, and then deciding what to do. And you're saying all of this could be turned over to agents. Layers of agents can handle all of this now, right?

That's right. So, now you have a layer, but you have feedback. You have a feed at each of these layers. There's a human being going, at the interpretation layer, "Do I think this is okay?" Then they hit a button and let it go to the next level.

So, it's an approval process.

An approval process, and also senior people looking over. They could be looking at agents looking at 6 different strategic options, right? Whereas in a very manual operation, that may take months to evaluate the competitive alternative. Now you're doing it in hours and days, right? So, that's the impedance mismatch there.

By the way, what we've seen historically is the impedance mismatch between a Fortune 500 company and a startup. The Fortune 500 company, to use it as an example, has so much to lose if they screw up that they're paralyzed in making decisions, and the startup is like, "Screw it. Let's just try everything."

Exactly. And this is just taking it one step further. In a big company, 1 of 20 people can say no to an idea and kill it. Whereas the startup can go to 1 of 20 investors, and 1 says yes, and they're off to the races. So, how do you balance that out?

Okay. So, now you have these layers of agents: purpose agents, sensing agents, interpretation agents, and decision agents. Our next level is an orchestration agent.

Let's say the decision agent comes back and says we should buy a startup that's doing this, right? Then the orchestration is saying, "Okay, we've got to go and set up a set of functions to go find a bunch of startups, analyze which ones are ready for M&A, tell the corporate development team, get the lawyers ready," et cetera.

And then get the legal agents ready.

Get the legal agents ready. And then finally, a learning loop: Where did we buy another company before, and did it work out or not? How did that work out? And all wrapped up in this governance thing.

That’s the kind of example of how you would flow through these. At the core is this engine: recursive learning. Another way to think about the organizational singularity is when you can have recursive self-improvement at the workflow level.

I love that.

So, if you took invoice processing, right now you have all these human checkpoints: Did the goods arrive? Who’s the supplier? Does the supplier exist in our systems? Is there a legal contract? There’s a human checking all those things. Maybe you have an ERP system that’s automated one or two of these layers, but now you can have the whole thing done, and then an agent can say, “How do I make this better every loop? How do I make this better at every loop?” It constantly improves.

Once you get to that level, you can actually step back, and you’re off to the races because everything should just self-improve at that level.

Yeah.

So, that’s the very heart of the whole thing with this layer. We also recognize that agents are going to be doing very crazy things, so how do you navigate that? We’ve come up with a framing that we found in smart contracts in Web3, plus some old web architecture, which says every agent should get a passport with a little metadata on what that agent is allowed to do or not allowed to do.

For example, policy-controlled APIs, object data, and object metadata that goes with it to say what that data is allowed to be exposed to or not exposed to. A liability framework is making sure your agents aren’t doing illegal things, because your lawyers will go bananas if agents go off outside your organization doing things and you have no idea what they’re doing.

So, every agent gets almost like a little passport on what they’re allowed to do. It’s constraints, oversight, and now you have other agents in the governance loop watching these things. The minute something goes off the rails, a human gets notified, the agent gets stopped, rolled back, checked again, and you can do that.

The reason this works is, in the quantum world, you need 1,000 physical qubits to make a logical qubit, right? Agents are relatively free, so you can have a lot of agents doing things, with a lot of agents overseeing them. So, the overall cost—you still get the benefits of that overall stack.

Here’s the question I’ll come back to for every CEO and every business leader out there: Could a 2- or 3-person team with Hermes or OpenClaw disrupt major lines of business in your business? If that’s the case, there are a few moats that you could develop.

One is proprietary data. That’s a clear moat if you have key data that can’t be replicated elsewhere. Number two is regulatory, which we see in health care and elsewhere.

Regulatory capture more than anything else.

And that moat can be eroded over time.

It can be.

All of these can be, but they’ll serve as moats for the time being. The biggest moat is an intelligence moat: If you can learn faster than everybody else, nobody’s going to catch you, right? This is why Claude and ChatGPT have learning loops further ahead than, say, Manus or Grok, and we’re seeing how quickly they’re moving ahead. Once you hit that, it’s very hard to catch up.

Another one would be really deeply committed to purpose and not wavering from that, because nothing shakes you in your relationship with the end customer and developing the depth there.

Yeah, a co-dedicated customer relationship, which feeds into proprietary data, right?

And brand. Brand is very critical because brand sits with MTP—that emotional connection with the end user. If you have a strong brand, you should use all of these new agents and capabilities to reinforce that. It means it’s hard to shake you out of that position.

Welcome to the health section of Moonshots brought to you by Fountain Life. You know, my mission is to help you use the latest technologies including AI to not just do your work at home, teach your kids, but to help you live a long and healthy life. I'm here today with an extraordinary physician, the chief medical officer of Fountain Life, Dr. Don Mussellem. Don, let's talk about cancer. You know, I know from the member database that we've have at Fountain, our members who come in who think they're healthy, it turns out 3.3% of them have a cancer in their body they don't know about.

Don Mussellem

That's right. You know, the majority of cancers that we screen for, those aren't the ones that are necessarily taking the lives when found at a late stage. We know that when cancer is found early, the chances for cure are much higher. We know it's much easier to treat a cancer when found early versus when found late. What we're finding in our members is over 3.3% were found to have these cancers that were otherwise wouldn't have been found or detected.

Yeah, you know, it's interesting, people you don't feel the cancer until stage three or stage four. And if you don't know what's going on inside your body, it's like driving your car with your eyes closed and you can know. And so when members come through Fountain, how do they detect cancers?

Don Mussellem

So we're doing full body MRI and we also do early cancer detection screening. This is very very important and these are not typical tools used in the conventional care setting when it comes to prevention. This is a hard thing because currently these are not studies that insurance would yet be covering but the goal is to collect these numbers, do the research and work hard to democratize wellness.

Yeah. So at the day you can know what's going on inside your body. It's your obligation to know. So check out Fountain Life. You can go to fountainlife.com/peter to get access to the latest technology to help you detect cancer at the very beginning at stage one when it is curable before it gets to stage three or stage four and you're world of hurt. So, those are some of the parts of that. Now let’s talk a little bit about what happens to the company and the classic organization, which has the C-suite, middle management, and the coal face doing things. What happens to them? C-suite—you already gave the example. What happens to them in this new world?

If you restructure—

Yeah. Have you given a name to the restructured organization?

ExO 3.0 is the best name I have. If anybody has a better name—

Okay. We would love to hear it. So, if you’re going from a classic organization, or an ExO 2.0, to an ExO 3.0, what happens to your organizational structure?

So, C-suite becomes basically accountability holders, dashboard oversight, evaluators, and validators rather than doers. You’re not going to be doing a strategic evaluation. Agents will do that. You basically hit yes, I like the evaluation, or not.

So, basically, you’re using your wisdom and experience to decide whether the agent’s action is in line.

That’s right. Now, this opens up all other questions, which we’ll get to in a second. C-level is guiding, holding accountability, watching what the agents are doing, and then deciding yes or no—do this, do that, whatever.

Middle management is where the biggest change happens, because middle management in existing companies is almost completely doing coordination. They take data from the coal face, repackage it for proper absorption by the C-suite, right? That function drops about 90%.

Mhm.

Then you need to lift up the human beings there and have them doing that exception handling, problem-solving, and so on, of which there’s a ton. We just don’t do it because most people don’t have time. Now you’ll have more time to do those things.

The bottom 20% are doing much more enabled work because agents are doing almost everything, and they’re also doing oversight and watching.

Now, we’ve talked about on Moonshots a number of times the idea that we’re going to see a reduction in the size of firms, from 100% down to 20%—an 80% reduction. You still—

The calculation is you’ll be able to run an average company with about 20% or 25% of the workforce that you had before.

Okay. Now, you can go down the negative side there, the immediate side, and go, “Oh, my God, 75% unemployment.” Or our Moonshots view would be that we’ll have 5 times more companies being created, and they’ll be that much more innovative.

The blossoming of entrepreneurship.

That’s right. We’ve seen the Cambrian explosion of startups already. We’re actually seeing hiring go up right now for entry-level jobs, which is really pretty interesting to spot.

Mhm. Okay, so those are the 3 things that happen to the 3 layers of the business. The question then becomes: How do you turn into one of these? And, by the way, where do you see the 80% being lost? At all of the levels, or mostly the middle levels?

No, I think 60% would be coming from middle management, 20% from the bottom, and 20% from the top.

Okay. That’s the full compression.

But mostly from middle management, because you don’t need to be gathering and aggregating sales reports. There’s no way you’re going to outperform an agent doing that. There’s much more work that needs to be done in the company that you could do that’s more valuable, right?

Now, an interesting question that comes up is the alignment problem: If you don’t have entry-level people doing the work, sweating it out, putting spreadsheets together, and doing the groundwork, what happens to your organizational and institutional knowledge?

Yeah, that’s right.

And where do you get senior management eventually when lower management and entry-level are not there? What we think will need to happen is very active and aggressive apprenticeship programs.

Mhm.

So, if you're suddenly a middle manager who gets displaced, go partner with the chief CFO and work on looking at alternatives. You'll learn a ton more, and you'll have a lot more fun. Back to the apprentice—really back to the apprentice, the guild kind of models. We think that'll start the thing.

Okay, so you have this new entity, this intelligence core, this new shape for the organization: C-suite, middle management, coal face. The next question really becomes: How do you get there?

Right. And this is the part where we have deep expertise, because when we built the ExO model, we decided one of the key things we had to solve was breaking that immune-system problem. When you try anything disruptive in a big company, the antibodies attack you.

Just to clarify this: When we say, “How do you get there?” how do you go from a classic organization to retooling yourself as an ExO Level 3? You're a $100 million trucking company, right? And now 2 guys can lease trucks, have an AI-centric organization, and compete the hell out of you. What are you going to do?

Okay, now this is the question: What do you do now, and how do you turn into this new model? What you do—and I cannot stress this enough, with the experience we've had—is you cannot change, fix, and transform the existing company.

It goes all the way back to Buckminster Fuller, who said you can't fix an existing system. You have to build a new system at the edge and let that become the new gravity center. John Hagel and John Seely Brown identified this as disruptive things happening at the edge.

The poster child here is Nestlé creating Nespresso in 1976. For 10 years, they tried to run it as a line of business inside the mothership. It didn't fit: different brand, different supply chain, different delivery, different customer proposition. Finally, they said, “Put it over there. There's too much friction inside the company.” They gave it to a different building, and boom.

Well, we wrote about this. The classic was Steve Jobs starting the Mac, or IBM creating its PC.

Or Lockheed with their Skunk Works, yeah.

Yeah. Apple would take a small team, put them at the edge, keep them a secret, and say, “Go disrupt a different industry.” Nestlé is a poster child of this. Nespresso is now one of their highest-performing lines of business, and every hotel room in the world has one.

So we know this. We've been talking about this for a long time with the ExO. We do disruptive things at the edge, and we've been working with Procter & Gamble, Siemens Energy, Black & Decker, and HP, helping them do disruptive edge innovation.

It's the human ego, in the final result, protecting itself from disruption. So, you have to do that different stuff at the edge. There's a reason why Amazon Web Services wasn't done in the core service; it just doesn't fit.

Yeah. Right? Okay, so you have to take this methodology and this approach. Just believe that.

You can try the other way. By the way, I tried in one of my companies—I’m not going to say which—in a 100-person organization, all right? I'm very much a compelling individual.

And I still could not get it, so I literally had to start it as a separate organization.

You do.

Yeah, and I've done that now multiple times.

Yes. And you maybe take it to an extreme, because every time something happens, you just spin off another company, which may be the Richard Branson approach. Every time he got to 150 people, he'd spin off another company to break through the Dunbar number problem, right?

But all I'm going to ask the viewers and listeners of this is to go research this to death. If you do anything other than disruptive things at the edge, pointing into adjacent spaces in a different way, you will fail. I've seen the innovation process in detail in probably 250 out of the Fortune 500, and I've never, ever, ever, ever seen any other method work than this.

I want to say one other thing: If you're going to try and do this on the edge, ultimately the edge organization needs to report into the CEO.

Yes, at the very top.

It cannot report in—

Another thing: The board of directors—

—needs to provide the CEO full support.

Yes. If you're disrupting your own organization and you don't have the board's support, you're screwed.

Yes. So, let me talk through how you do this.

Yeah. You do not touch the existing organization. It's your revenue engine.

Yeah, don't touch the cash cow. What's happening right now is people are trying to stick AI into places, and it's just not working.

What you do is, at the edge of your organization, you create an AI-native digital twin. Then, once you set up that separate entity, take 3 to 5 of your crazy young people and partner with a company that's a builder—not a consulting company, but a builder—so you get what's called forward-deployed engineers, which is the latest buzzword in software these days.

What you do is pick a workflow. You've got all these workflows in the legacy organization. Is that a product or a service? Well, call it invoice processing. It's a workflow. That's a very standardized, cookie-cutter workflow where you know exactly how it works.

You rebuild it in this new entity. You don't move it; you copy it. You take the steps in this workflow, and we've got a whole methodology for how to break down and score each task. That's built into the methodology of the whole approach.

You replicate it in this new system. You fork the data so that you have the data to do it, and now you start running it here. You've de-risked it, also, because if something goes horribly wrong, you're not risking the mothership. I cannot stress this enough.

So, you run this in parallel until you hit that recursive self-improvement loop. Once you see that the improvement loops here are way faster than you can achieve them here, then you know you're onto something. Even then, give it another few weeks to quality-check against the original.

Quality-check against the original, yeah.

Quality-check that you've got everything, and then you slowly deprecate the old system and take on the next workflow. Maybe it's receipt confirmation, and you move that over. Maybe the next one is demand forecasting, and you move that one over. Little by little, you grow this thing at the edge—a full digital twin.

And a full digital twin—

—that's in recursive self-improvement.

The next thing you know, you've got your AI-native digital twin fully running. Our current estimates are that once you have that digital twin running properly, your performance improvement should be 100X or higher per year. Just 100X better. If it's processing 1 invoice now, it should process 100 invoices next. If you were taking 100 days to do something, it should take 1 day to do it.

What's the scaffolding around the digital twin?

Well, that's the whole thing. That's where you're building up this system. In this new model, you have human beings there, but there are fewer of them, and they're doing more oversight, exception handling, problem-solving, and so on. You're literally building your AI-native digital twin at the edge.

Okay. What gets me excited as well is the idea that once you've done that, you can start to create adjacent companies. You can spin off anything. If you're a great entrepreneurial team and you're limited by—I have a lot of companies with amazing teams of people doing things, and I don't want to push them any further because of quality of life; they'll break, they'll get stressed out—but if all of a sudden you can get that automatic digital twin running, that team can now start building other products and services.

Exactly. You can do that. Now, let me give you a real example. There are 2 sectors, by the way, that have gone through this full loop. One is contact centers.

We used to do human business-process outsourcing. We had call centers doing stuff. Then phase 2 of that automation was chatbot-assisted customer service. Now we have AI-native customer service. Klarna has done this.

Just talking to the AIs on X. Yeah, it's all Grok-driven.

It's all Grok-driven, right? I set up a new website for the Organizational Singularity, and I went on Cloudflare. The AI told me exactly how to run the exception rules for domain forwarding. I was like, “This is incredible.”

The automation that's going to be possible is going to be magical to people. Anyone who's using AI at a refined level today sees how much fun it is compared to what it was like before. We've seen this to the point that we're working a 7-day week.

We're killing ourselves.

But everybody's having so much fun now. It doesn't feel like work.

No, it's play because we're getting so much done. It took 3 years of hell to write the first book, right? It took us 2.5 years of hell to write the second book, mostly because we had to rewrite it with—

Because we had to deal with me.

No, no, no, no, no. We had to rewrite it because generative AI came out near the end. But this third book took 3 months, right? Every contributor could use AI, add more data and help to it, add their methodology to it, and then, boom, you're off to the races.

The second domain where this is fully happening, by the way, is marketing and content generation, right? We used to have an agency-heavy model, then it was AI-assisted, and now it's AI-native. We can see certain verticals hitting this spot in a particular way.

Let me go into the rewriting methodology. We call this methodology REWRITE. I want to go into a little bit of detail so people understand the specific steps that are involved. You have a workflow, like invoice processing, and you're going to start moving the workflow over. Before you do any of that, you have to do a backcasting exercise.

What's that mean?

Backcasting is a methodology in future studies and forecasting where you pick what the vision looks like. Say Elon wants to get to Mars. You could say, "Okay, I want to get to Mars in 7 years. In order to get to Mars in 7 years, where do I have to be in 5 years? Where do I have to be in 3 years?"

Sure.

And now you have your roadmap. If you start from the starting point and say, "I want to get to Mars," you have no idea where you're going or how you're going to get there. Backcasting has turned into a very powerful methodology.

Step one is to take your company—let's say it's that trucking company or retail company that I used earlier—and say, "Okay, in this future world, what does that company look like, fulfilling its MTP and its architecture in an AI-native, AI-centric way?" Then you paint that picture.

One of the hardest things for people to do is to let go of how they've done it.

Yes. And, by the way, it's also one of the easiest things to do in conversation with a large language model. Beautiful, right? Take your C-suite and go do that backcasting exercise. That's phase one. We have people who can help people do that.

Step two: you score your company. We've got a whole bunch of metrics on which we want to score the existing organization. For example, I'll just give you 2 of them. One is: What is the organizational drag inside your organization? Right now, if you try to get something done, does it have to go through 5 or 6 different decision loops and approvals before you get it done, or can they, like at NVIDIA, go straight to the founder and say, "Can I do this?" and he says yes or no? Or an AI tells you whether you can do it or not.

So, what's the organizational drag on a scale of 1 to 10? A second metric would be: Where is AI a first-class citizen in your company right now? If it's a tool injected by IT, you're on the lower end of the score. If you've got a chief AI officer and you're building AI-native capability already, your score is much higher on that 1-to-10 score.

You've got 7 dimensions. We ask you those 7 questions and you score yourself. We'll have this on the website for people to take for free. Evaluate yourself. It's a 1-to-7 thing.

The next step is to take the most prescriptive workflows you have in your organization and start mapping and documenting them so you have clear knowledge. A big problem, by the way, is going to be what's called tacit knowledge. Let's say you're doing video production. There are a bunch of steps you're doing as a video producer that may not be obvious from the outside. They're not documented anywhere. If you lose that person, an AI can't do them right away. It's the unspoken knowledge.

And, by the way, there's a whole process right now by which companies are basically shadowing you with an agent.

They are trying. They are trying to shadow, yeah.

But it turns out, if you're a Gen Z worker, 44% of Gen Z workers are sabotaging the AI and giving it bad information so it can't take their job later.

Wow. It's that level of immune-system response, right?

By the way, that is a perfect example of the immune system.

You're trying to do something, but the culture is killing you as you're trying to get that done, right?

I'm going to reiterate this. We've created a 10-week process where we've found a way of hacking and breaking the immune system, hacking culture at scale. We've done it 100 times for big companies.

I love it. I played in a little bit of that, and I love it.

The next step is to cut the organizational drag. Start stripping out approval levels in your company so that you actually strip things down to, "You can break it." What would that look like? Okay?

The next step is to start building that digital twin and migrating workflows over one by one. The final one is to rewire your systems more and more so that everything is going to that rather than to this.

Let me take one more crack at visualizing this. Today, this is how most companies operate. They have their cloud provider, their networking, and their other capabilities. Then they have a set of ERP systems—Oracle Financials, SAP, whatever. All the data sits inside those systems, right? Those companies don't want you to have that data easily, so it's wired in.

Then you have an application layer, and people are trying to layer AI on top, hacking against this horrible architecture that we've had for 50 years. It can't be easily unwound.

Picture the new architecture. You've got connectivity to a cloud provider, a data lake that has all your data accessible in one spot, with the proper approval levels attached to each data object. Then you have your application layer, which is custom-built for you because AI can do that, and workflows, et cetera. Then you have your AI, and then your agents on top of that. This is a wholly different stack and architecture.

That you own. You own completely, right?

Yeah. And this is why the SaaS providers are so freaked out, because that model is not compatible with this model, right? Right now, they're trying their best to keep their place because they're wired into the operating system of the legacy organization. But if you build this proper stack, you have full agency and control at a much cheaper cost than you could have had before.

Ask anybody who's tried to implement an ERP system how much hell they had trying to do it. Then you end up trying to map the organizational flow to the ERP system, rather than the other way around. Now you can have software built that way. We've built a whole methodology for this.

The last couple of points around this: We think this overall transition is going to take about 5 to 7 years to complete.

Wait, wait. Let me fully understand it. Not for a single company to do it, but for all companies to get there?

For the majority of companies to get there.

For the surviving majority of companies, over a 5-to-7-year period, you're either dead or you've transitioned to this.

This maps as well, by the way, to the conversation we've had about the turbulent period of time. We actually call this the turbulent transition, exactly for that reason.

You said it's 2 to 8 years. We have to carefully architect, societally, how we get through this 2-to-8-year period.

Yes.

And I'm just talking about companies. Forget anything else.

Yeah, but it's the underlying reason.

That's right. Okay.

So, in our opinion, you should be able to run a company with 10% to 25% of the people that you have today.

You mean just—

Well, if you're regulatory-centric or have physical work, like if you're building a data center, then it's less. If you're a marketing company, then you're going to be down to 10% human beings, right? But if it's a physical company, even then it's only 25%.

For example, we were doing work with Fermi America, and we estimated that we should be able to run a power plant with about 80 people instead of 800 people. That's a full 10% headcount there.

There should be 1 manager for every 20-plus high-impact individual contributors, what Jack Dorsey called HIIC, instead of 1 to 5 or 1 to 3, as it is today.

He took it to an extreme, all right? He wanted to have just the CEO, with everybody connecting to him.

But what that means is he's using AI to do everything, because there's no way that the CEO can keep that many people connected to that many people anyway, right?

This is already happening. Take Cognition Labs. Their ARR grew 73 times when they implemented this full system, when they went fully AI-native. This is already happening.

Yeah. This is not some pie-in-the-sky guess.

We're taking early signals, and over the last few months, as we've been watching the market evolve, every single data point we've gathered is pointing exactly at the trajectory we're pointing at.

This is actually a race. If you're a company in an industry and someone else runs this process and has a recursively improving digital twin—

Yes.

—and you don't, you're cooked. You're cooked.

Yeah. That's right.

So, if you're Unilever and Procter & Gamble is taking all their stuff and automating it, you will not outcompete them, right? Or the other way around, right? Whichever way it is.

Okay. So, let me talk about what survives and what doesn't survive today.

By the way, just to hit it, our friend of the pod, Elon, has talked about increasing GDP—triple-digit growth. I mean, this just adds rocket fuel.

It's insane.

Yes. We're going to see insane levels of growth of companies that are delivering 100x compared to what was being done before. We're doing things 100x cheaper.

In terms of profitability, right?

That's right. Your revenue scale and profit go through the roof.

Yeah. Now, profitability will be limited because, with that profit margin, other companies are going to go, “Wow, look at that profit margin. I'm going to send my AI agents to do that.”

Which is why things demonetize and why we end up heading toward universal high income, because the cost of everything starts dropping down. Then we can get into the whole UBI, UHI, universal basic services stuff, et cetera.

This episode is brought to you by Blitsy, autonomous software development with infinite code context. Blitsy uses thousands of specialized AI agents that think for hours to understand enterprise-scale codebases with millions of lines of code. Engineers start every development sprint with the Blitsy platform, bringing in their development requirements. The Blitsy platform provides a plan, then generates and precompiles code for each task. Blitsy delivers 80% or more of the development work autonomously, while providing a guide for the final 20% of human development work required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitsy as their pre-IDE development tool, pairing it with their coding co-pilot of choice to bring an AI-native SDLC into their org. Ready to 5x your engineering velocity? Visit blitsy.com to schedule a demo and start building with Blitsy today.

All right, let me just do a before and after.

Sure. So, what survives is what the new entity looks like: the MTP encoded as a protocol in the company. Number two, the accountability shell: legal entity, fiduciary holder, liability container, et cetera. Proprietary intelligence in that stack is very, very critical.

Coordination protocols become very killer. Curatorial judgment: when execution is nearly free, judgment and taste become really important in the future.

Yeah, we've talked about that. Super important.

Those are the things that will survive and thrive in this new world. What does not survive? Number one, the org chart, the way we built it. David Rose is famous for saying, “The org structure that got you successful in the 20th century will have you fail in the 21st century.”

Yes. Turns out he was right. It's just taking a little longer.

Iterate it again. The org chart in the traditional model completely fails. The 5-year plan dies completely. In fact, any static planning dies, because if you do any strategic thinking of, “This is what the world is going to look like a year from now,” you have no concept.

We need constant learning loops. We're in the middle of the singularity. You can't rely on any static plan.

Imagine people who are seeing this are getting very anxious right now.

When I've spoken about this at conferences, people are like, “My head is breaking, freaking out, dying here.” But again, it looks like we've found a very stable mechanism to get you from A to B.

So there's some comfort level. I think that's so important.

Well, in terms of what the world needs, this is what you and I love doing: stuff that the world needs. It's very clear this is what the world needs—a stable framework to get us from A to B.

Yeah. And if we can have a little less of the chaos as old systems fail, and we can fail over more elegantly, then let us, please—goodness—do that.

So, the 5-year plan—in fact, we actually took it to the point where, right now, if you have an organization, that org structure changes only when you have a major event, like an M&A transaction, or you launch a new line of business or something.

Or you replace the management team.

So, that org structure does not change very much. But in the new world, that org structure is dynamic and constantly changing, adapting to the current situation. It's like an amoeba. And that's the org structure.

Forget it. The organization itself becomes a protocol. That's a big thing to get your head around. Middle management as a coordination layer: gone. Quarterly reviews as a unit of decision-making: gone. Annual planning: gone.

Yeah. Inertia mode—customers don't switch because switching is annoying: gone. Wasting assets in the aging economy: gone.

So, there's a bunch of things we've highlighted about what happens first. One piece of guidance I would give to people is, if your company is less than 50 people, you can brute-force this and do it in the whole company because you've got a first-name basis with everybody.

If your company is over 50—in your case, it was 100—do not try to break the immune system. You'll risk the existing company, and you don't want to do that. Do this digital twin at the edge.

Yes.

So, what we're doing right now is saying, “Okay, let's pick a few CEOs that want to go through this.” We're going to score them in that rewrite score. We're right now at about 4 companies. We're kind of going through them with that. We'll probably do 10 at a time.

Let's be very specific about that, because I can imagine a lot of our viewers want this, and we have a lot of large entrepreneurial companies and so forth. If someone does want to be one of the first 10 going through this, who do they email? Where do they go?

Two paths: email kevin@openexo.com. Kevin Allen is our head of community and navigates all this. Or go to our website, organizationalsingularity.com, and you can fill out a form and say, “I want to try this.”

K-E-V-I-N at openexo.com.

Kevin@openexo.com, right? Or go to organizationalsingularity.com.

And you're going to selectively choose who you work with.

Selectively, yes. Let's say a company has horrible organizational drag. We're going to say, “Go fix the organizational drag first,” because we're going to spend all your time on that. We think it's a 90-day process.

To do what?

To run this process and get a few workflows working in this new way. Once we get you going, you should be off to the races, and you can build on yourself.

I will take batches. The first batch will be 10 or 20, probably, and then we may do more. We'll see how that goes. My entire community is being retrained for this. My ExO community is now 50,000 people in 150 countries, so we're retraining them to be able to navigate this.

We're all going to go through this journey together. I'll be personally involved in the first couple of batches, like I was personally involved in the first sprints, et cetera, to make sure this is happening.

I just heard Sheikh Mohammed say that he wants to run 50% of the Emirati government on this. Do you see this working for governments as well?

Completely. Think of any government. Almost all the processes in a government are prescriptive and very well understood. The process for renewing a driver's license is extremely well understood and frustrating, but now that friction can be removed in a really magical way.

In fact, they've done this. Minister Al Olama said, “Salim, come and get a golden visa. You're going to be my poster child.” And they're processing golden visas in 5 hours—a residence visa in 5 hours. This is unheard of in that world.

They've already been down a path like this. They're taking it naturally to a whole other level. But for governments and nonprofits, this completely applies. There's a whole chapter we have in the book, which I won't talk about here, but go into the whole “Solve Everything” paper that you and Alex did.

All of Alex's thinking on the inner loop—we've taken a crack at how to organize domain after domain, create a domain collapse in more and more sectors, and organize for that.

Yeah. So, you can create an organizational design where you can pick a domain like healthcare or education and set up a structure that then has that inner loop start to move. And I guess the other question is, if you're an entrepreneur thinking about starting a company—

Yes. You have basically a platform here and a playbook to start.

That's right—immediately.

Yes. Now you can read this. In fact, what we're going to do is launch the book as an API—as an AI. We're going to launch it as a Claude skill—

All right.

—that you can just download. Claude just said, “Hey, we're going to have connectors to all QuickBooks and everything else like that.” So, we're going to download the entire contents of the ExO framework as a Claude skill, because every 2 or 3 days we're learning new things.

We're going to build it in, so then the skill itself is changing on a real-time basis. You can't just get certified in this based on something from 5 years ago. The AI itself has to stay updated. So, we're releasing the book as an AI—

Nice.

—as a native AI.

Amazing. So, I guess the question is, if you're ready for this and you're selected, that's great.

If you're a company that's got too much—what do you call it? Organizational friction?

Yes, so organizational drag. So what do you do? Oh, come and see us, because we'll show you. We'll tell you what to do. For example, if you've got a process that takes 10 steps, brute-force it and rethink that process so it takes 3 steps.

Once it's taking 3 steps or less, then you're ready to start thinking about moving over into the digital twin. You can also start setting up the legal framework for the digital twin and get board approval. There's a lot of scaffolding that has to take place for you to get there.

You may have legacy legal issues. For example, in Germany, works councils decide how many employees a bigger company is allowed to have or not have, which is not great from a flexibility point of view. But there's so much else you can do to start navigating this. In fact, one of our folks, Patrick Sandino, said, "Look, let's figure out a way, in this process of retraining all of the people doing work that might be at risk, to retrain them to be in this new model so that you have a kind of whole transition plan for society built in." So you solve the social contract along the way. We'll see how that works out.

Amazing. Again, just to reiterate, if someone's interested, kevin@openexo.com.

Yep. Or go to organizationalsingularity.com.

No, I love this, Salim. You've been pregnant, giving birth to this for a while. We've been talking about this. It's about 3 months of stuff. What I did was start writing the first version of the book and work with Claude and ChatGPT on 3 instances of Gemini, ChatGPT, and Claude, each taking cracks at different things.

Then I sent it out to the community and said, "Give me feedback." We got lessons learned, and then we went and talked to some of the cutting-edge AI practitioners. What are you doing? What are you seeing at the cutting edge?

It's been a process, because the field is changing as fast as we're able to keep up with it. Just keeping up with this—like Moonshots, right?—we're spending a huge amount of time just keeping up with all the breakthroughs and headlines. We're having to have a team dedicated to just keeping track of all the things happening so we can constantly tweak the methodology itself on how to do the rebuilding.

I think this is teaching boards and founders how to survive the next disruptions that are coming. The disruptions are coming. The disruption is now. It's like William Gibson said: "The future's here; it's not evenly distributed." The organizational singularity is here. It's just not evenly distributed. If you're a 5-person startup, you're building an AI-native way anyway.

Yeah. We have a whole bunch of our community members who are doing that, and we've been learning from them. You see Alex Finn with all the OpenClaw stuff and the Hermes, and what that's making possible. The big, central thing to think about is that all of our organizational structures in the past were organized around hierarchy and human-centric workflows.

Now they need to be AI-native, agentic workflows, and that's a totally different model. It needs to be architected around intelligence, not around hierarchy.

Love it. I hope, on our weekly—soon biweekly, and soon daily—Moonshots—

Daily? Oh, my God, you're going to—

I know, it's crazy. But do you know how many flights I've had to change over the last few weeks? Oh, my God. The only flight I could take is right when Moonshots are happening. I have to now stay till the next day. How many airports have you broadcast from? It's been bad. It should get better, by the way. It should get better. But I hope that we'll be able to track this, and you can report on companies that have made this transition.

That's right, and how this is updated. For me, this is one of the most important learnings that you can deliver. I'll give you one early thing we've seen. You know what's one of the biggest categories of people that are approaching us? Universities. They're like, "We need to automate. We need to totally change. We can see the writing on the wall."

Yeah. Massive disruption coming.

So they're coming, going, "How? What do we do?" And we're like, "Great, let's start with you. Let's start automating the existing one and move you into this new model," so that as you turn from trying to teach content to teaching execution, you become entrepreneurial hubs. We talk about the fact that your engineering degree won't be that you studied engineering for 4 years. You built a bunch of stuff, and it was interesting enough that you got credentialed.

That will be the engineering degree. It'll be doing rather than learning. That's such a big shift for the legacy system. What I'm really impressed by is that they're seeing it. I didn't think they would even see it, but they're actually seeing it and reaching out to us.

Amazing. Listen, buddy, I thank you for sharing this. It was actually amazing to see your brilliance and your passion about this. To all of the community, it's your drive here. This is your heart and your soul. This is it. I mean, this is how you organize for the new world, right?

If you're going to rebuild civilization—rewrite civilization, right?—you have to think about how the organizational design around this all works. We have to rethink the whole thing.

If you're the CEO, this is coming. There's no ifs, ands, or buts about it. It's happening at an accelerating rate. And remember that disruption is not coming from your largest competitor. It's coming from the AI-native startup that sees how slow you are and how much profit you're currently making, and they're going to come and try to eat your lunch.

I think your T-shirt says it all: "Abundance is coming."

Yeah, abundance is coming.

Yes. Brother, thank you for this. It's beautiful, and I love spending time with you. I'm excited to go and celebrate your birthday tonight.

We will do that.

Yes, fantastic. If you made it to the end of this episode, which you obviously did, I consider you a moonshot mate. Every week my moonshot mates and I spend a lot of energy and time to really deliver you the news that matters. If you're a subscriber, thank you. If you're not a subscriber yet, please consider subscribing so you get the news as it comes out. I also want to invite you to join me on my weekly newsletter called Meta Trends. I have a research team. You may not know this, but we spend the entire week looking at the meta trends that are impacting your family, your company, your industry, your nation. And I put this into a 2-minute read every week. If you'd like to get access to the Meta Trends newsletter every week, go to dmandis.com/metatrends. That's dmandis.com/metatrends. Thank you again for joining us today. It's a blast for us to put this together every week.

Yes, fantastic.