Kanoa

OpenAI just offered the U.S. government a 5% equity stake in the company. A $42 billion position. And they framed it as sharing AI's benefits broadly with the American people. I want to sit with that framing for a second. Because it sounds generous. It sounds civic-minded. It sounds like the kind of thing you say when you want a headline that reads well and a regulatory conversation that goes smoothly. But the government is not the public. And I think that distinction matters more than almost anything else being said about this deal right now. Let's start with what's actually happening. OpenAI is in the middle of converting from a nonprofit structure to a for-profit entity. That transition has real implications for who benefits from the value the company creates. As part of that process, they floated the idea of offering the federal government an equity position. They even invoked the Alaska Permanent Fund as a comparison point — the idea being that a government-managed fund could distribute AI's returns broadly, the way Alaska distributes oil revenue to its residents. It's a clever analogy. And it's worth taking seriously before explaining why it doesn't hold. The Alaska Permanent Fund was established in 1976. The state legislature amended the Alaska constitution to require that at least 25% of oil revenues be saved rather than spent. Since 1982, every eligible Alaskan resident has received an annual dividend check from that fund. In 2023, that check was $1,312 per person. It's direct. It's individual. It's a genuine transfer of resource revenue to the people who live on top of the resource. There's something philosophically honest about that model. The oil belongs to Alaska. Alaskans own the land. The revenue flows to them. The chain of ownership is clear. Now let's map that onto what OpenAI is proposing. The federal government would own 5% of OpenAI. That government would then presumably manage that stake in the interest of the American public. Except the federal government is not Alaska's legislature managing a constitutionally protected fund. It's a $6 trillion annual spending machine with its own interests, its own contractors, its own political pressures, and its own definition of what serving the public means on any given Tuesday. When Alaska sends you a dividend check, you get the check. When the federal government owns equity in a company, you get whatever policy outcomes follow from that ownership — which might be nothing you ever see directly, filtered through budget negotiations, defense appropriations, agency priorities, and the general friction of a government that has twelve other things happening at the same time. Those are not the same thing. Not even close. But there's a deeper problem here, and it's structural. OpenAI didn't build its models in a vacuum. GPT-4, GPT-4o, the entire product line — these systems were trained on human expression at a scale that is genuinely difficult to comprehend. Books, articles, forum posts, social media threads, code repositories, personal essays, comment sections. The raw material of these models is the accumulated written output of millions of individual people who had no idea their words were going to be used this way, who consented to nothing, and who will receive no return on that contribution. The individual was the input. The institution is capturing the output. That gap — between who provided the raw material and who captures the financial return — is the actual story. And it doesn't get better by adding the federal government to the cap table. If anything, it gets more opaque. Now you have a trillion-dollar AI company, its institutional investors, and the United States government all sitting between the individual whose expression trained the model and any possible return on that expression. Historically, this pattern is familiar. It shows up whenever a new resource class gets discovered and the question of ownership gets settled before the people closest to the resource have any real leverage. Oil in the early twentieth century. Radio spectrum in the 1920s. The internet's data economy in the 2010s. Every time, the framing is that broad access to the product is the same as ownership of the underlying asset. You can use Google Maps, therefore you share in the value of location data. You can use ChatGPT, therefore you share in the value of the model. That's not ownership. That's consumption. The Alaska model is actually instructive here, but not in the way OpenAI intends. What made Alaska's approach meaningful wasn't that the government managed a fund. It was that the ownership right was explicit, individual, and constitutionally protected. Every Alaskan has a legal claim to a share of the revenue. That claim doesn't depend on legislative goodwill in any given year. It's structural. Real individual AI ownership would look something like that. Not a government stake. Not a sovereign wealth fund managed by appointees. It would look like individual data rights that carry economic weight — a legal framework that recognizes your expression as a contribution to the model and entitles you to a proportional return. It would look like equity structures designed to distribute ownership at the individual level, the way some cooperative models have attempted to do in other industries. It would look like governance rights attached to that ownership, so the people whose data trained the system have actual standing when decisions are made about how the system is used. None of that is easy to build. The coordination problems are real. The valuation problems are real. Figuring out how to attribute a specific individual's contribution to a model trained on hundreds of billions of tokens is genuinely hard. But hard is not the same as impossible. And the fact that individual ownership is technically complicated doesn't mean that a government equity stake is a reasonable substitute. Those are different problems. One is engineering and legal architecture. The other is political optics. I'm not saying the people at OpenAI are acting in bad faith. I don't think the calculus here is simply cynical. But I do think incentive structures matter more than intentions, and the incentive structure of this deal is straightforward. OpenAI needs regulatory goodwill. Offering the government a $42 billion stake is a very efficient way to align the government's financial interests with the company's survival. That's not public benefit. That's regulatory capture with a civic wrapper. The framing of this as an Alaska Permanent Fund moment depends on a confusion between the government and the people it governs. They're related. They're not the same. A government stake in OpenAI means the government has more reason to want OpenAI to succeed. It does not mean you do. If AI is really for everyone — if that's the actual claim, not just the marketing — then the ownership structure needs to reflect that. Not eventually. Now, while the structures are still being built. Because the decisions made during a technology's formative period tend to calcify. The early radio spectrum decisions shaped broadcast media for a century. The early internet infrastructure decisions shaped who captured value from the data economy for decades. We are in that moment right now for AI. And the current answer to the question of individual ownership is: the individual was the raw material. They're still not the owner. That's worth saying clearly, even if it makes the press release harder to write.

There's a sentence I keep coming back to. It's not from a manifesto or a policy paper. It's just a quiet fact that nobody in the industry wants to say out loud. You were never the customer. You were always the input. Let me explain what I mean by that. When OpenAI started building its own chips, the story was efficiency. When they acquired Jony Ive's design firm, the story was beautiful hardware. When they pushed deeper into the consumer interface layer, the story was seamless experience. But zoom out and the picture is different. What you're watching is a company building a closed stack from silicon to screen. Your thoughts, your questions, your work — all of it runs on their infrastructure, gets processed by their model, and flows through their interface. Every layer is theirs. You contribute the most irreplaceable part of the system — the human reasoning — and you own none of it. That's not a conspiracy. It's just the logic of vertical integration. Control the stack, capture the value. So people started pointing at open source as the answer. And I get it. The instinct is right. If the weights are public, if anyone can run the model, then nobody owns the intelligence layer. That argument makes sense in theory. Here's where it breaks down. 99% of people using open source AI are not running models locally. They're using Groq running Llama. They're using Perplexity running open weights. They're on hosted platforms where a third party's data policy applies the moment they send a message. Open weights and data ownership are completely separate properties. The model being open does not mean your conversation is private. It does not mean your data stays with you. Most people don't know this distinction exists, and the industry has no real incentive to explain it clearly. Truly local open source — running a model on your own machine, your own hardware, air-gapped from anyone else's servers — is the most sovereign setup that exists right now. It's also almost nobody's setup. So we have closed stacks capturing your data by design, and open stacks capturing your data by default. Two different architectures, same outcome for the individual. But here's what I think is the part nobody's talking about seriously yet. The weights are not the most valuable thing in this system. You are. Not you as a demographic. Not you as a behavioral signal. You as a specific person with a specific history of thought. Your conversation logs. Your reasoning patterns. The way you frame problems. The expertise you've built over years that shows up in how you ask questions and what you notice in the answers. That dataset — your individual cognitive fingerprint — is more valuable than any open weight model. It's what makes AI genuinely useful instead of generically useful. And nobody at the frontier is building with the individual as the owner of that dataset. Not really. The incentive structures don't point there. The investment theses don't point there. The product roadmaps don't point there. That's the gap. I'm not here to tell you the closed labs are evil or that open source is a lie. The people building these systems are making rational decisions inside the incentive structures they're in. I'm just trying to name the structure clearly, without the tribalism that usually surrounds this conversation. The question isn't which AI company you trust. The question is whether trust should even be the mechanism. And I don't think it should.

The AI stack is splitting in two. And most people haven't noticed yet. On one side, you have the closed labs. OpenAI isn't just building models anymore. They're building chips. They're acquiring infrastructure. They're locking down every layer of the stack from silicon to interface. The goal isn't just to win the model race. It's to own the entire surface area of AI interaction. Every query, every session, every pattern. Vertical integration at a scale we haven't seen since the early internet platform wars. On the other side, something different is happening. Labs in Japan, China, and Europe are releasing frontier-class models that anyone can download and run. The open source movement in AI has gone from scrappy to serious. These aren't hobbyist projects. These are models that can compete with the best closed systems in the world. The idea that cutting-edge AI requires a closed proprietary lab is no longer true. So open source won, right? The model problem is solved. Not quite. Here's the thing nobody is talking about. You can download the weights. You can run the model on your own hardware. You can fork it, fine-tune it, ship it. But you still don't own your data. Your conversations. Your behavioral patterns. Your identity as expressed through every interaction you've ever had with an AI system. That layer doesn't move when the model does. The model is the engine. The data is the fuel. And right now, the fuel stays with whoever built the car. This is the unresolved gap in the entire AI ownership debate. Everyone is arguing about which model is best, which lab is winning, whether open source can keep up with the closed frontier. Those are real questions. But they're the wrong frame. Because the model is not where the value is captured. The data is. Think about what your AI interactions actually contain. Your questions reveal your priorities. Your follow-ups reveal your thinking style. The topics you return to reveal what you're working on, what you're worried about, what you care about. Over time, that isn't just usage data. It's a portrait. A behavioral fingerprint that is more accurate and more intimate than almost anything else that exists about you. And right now, that portrait belongs to someone else. The individual was never just a user in this system. They were always the raw material. The training signal. The behavioral dataset that makes these models more useful, more personalized, more sticky. Every interaction you had was a contribution. You just didn't get equity. That's the problem Uare.ai is built to solve. Not the model problem. The ownership problem. The question isn't which model runs your AI. The question is who holds the layer that makes your AI yours. Open source cracked the model open. That matters. But the data layer is still captured. And until that changes, the power dynamic doesn't change. You're still the raw material. You're still producing value you don't own. The question now is whether the individual finally becomes the owner. And what it actually takes to get there.

Everyone laughs at Myspace. The glittery backgrounds. The auto-play music. The chaotic HTML profiles that broke your browser. People look back at it like it was a mistake. I think they're completely wrong. Myspace had the right idea. Total personal customization of your corner of the internet. Your profile was yours. It looked like you, sounded like you, felt like you. That vision was correct. The problem wasn't the vision. The problem was the friction. To make your Myspace page actually yours, you had to learn HTML. Most people couldn't do that. So the platform that promised individuality ended up being a place where only the technically brave got to express themselves fully. Everyone else got the default. And eventually, Facebook showed up with a clean, uniform layout, and the world picked simplicity over self-expression. But here's what actually happened: we didn't choose conformity because we wanted it. We chose it because individuality was too hard. That's the part that stuck with me. Fast forward to today. The SpaceX IPO just landed at a $1.77 trillion valuation. That number is worth sitting with. Smart money is pouring into infrastructure. Not applications. Infrastructure. And the reason is simple: nobody knows who wins the application layer yet. The race is still wide open. The people writing the big checks know that whoever controls the rails wins regardless of which train ends up being the fastest. But here's what I keep coming back to. Every AI application being built right now is still Web 2 in its soul. You open the app. You learn its interface. You adapt to its logic. You conform to the product. The tool doesn't know you. It doesn't remember how you think, how you communicate, what you care about, or how you make decisions. You're still doing the HTML. The friction just looks different now. AI changes that. For the first time in the history of technology, the tool can actually morph to the individual. Not the other way around. Your AI can learn your voice, your reasoning style, your expertise, your preferences. It can become a truer version of how you think, not a generic assistant that treats every user the same. That's what Myspace was reaching for and couldn't grab. The vision was always right. The technology just wasn't ready. It is now. That's exactly what we're building at Uare.ai. An Individual AI platform. Not personal AI in the watered-down sense the industry throws around. Individual. Your AI molds to you. Your preferences, your expertise, your identity. The platform learns from you over time and reflects you back with higher fidelity the longer you use it. The winner of this race won't be the company with the best model or the most integrations. It'll be the company that commits fully to user individuality. That's the bet we're making. If you want to go deeper on any of this, you can chat with my AI directly at Uare.ai. Any feature requests or feedback you share goes straight to our team and directly shapes what we build next. That's not a marketing line. It's how we actually operate.

There's a question nobody in tech wants to sit with for too long. If you automate 80% of your job, are you still doing the same job? I don't mean that rhetorically. I mean it as a real philosophical problem. The Ship of Theseus says that if you replace every plank of a ship, one by one, the thing you end up with both is and isn't the original ship anymore. It just looks like it. It floats the same. It sails the same. But something essential has been swapped out. Knowledge work is going through that right now. And almost nobody is talking about the part that actually matters. The conversation we keep having is about productivity. How much time AI saves. How many tasks it can handle. How many headcount decisions it justifies. That conversation is real, but it's also shallow. It's measuring the wrong thing. Here's what I've watched happen, up close, at an AI startup where this isn't theoretical. When repetitive cognitive work disappears — the summarizing, the drafting, the scheduling, the sorting, the formatting — something unexpected fills the space. People stop talking about what they do. They start talking about what they think. The identity doesn't follow the task anymore. It follows the judgment. That's not a small shift. That's a complete restructuring of how someone understands their own value. I'm a Chief of Staff. A year ago, a meaningful portion of my time went to work that required cognition but not much judgment. Synthesizing notes. Drafting comms. Tracking action items across threads. I was good at it. I was fast. And I would have told you, if you asked, that being good at those things was part of what made me good at my job. I don't think that anymore. Not because the tasks weren't real, but because once they were handled differently, I could finally see what the job actually was underneath them. The job was never the tasks. The tasks were just the medium the job happened to travel through. This is where it gets uncomfortable. Because if that's true for me, it's true for a lot of people. And that means a lot of people have been identifying with the medium instead of the message for their entire careers. Not because they were wrong or unambitious. Because that's what the environment rewarded. Speed. Output. Volume. The ability to move more planks than the person next to you. AI didn't just automate the tasks. It exposed the question we never had to answer before: what are you actually here to contribute, when the execution is taken care of? Some people find that question exciting. A lot of people find it destabilizing. Both reactions make complete sense. Because here's the thing about the Ship of Theseus that usually gets lost in the philosophy seminar version of the argument. The ship doesn't get to have an opinion about which planks get replaced. It just floats or it doesn't. We're not ships. We get to have an opinion. We get to decide whether the version of us that emerges from this transition is truer to what we actually are, or just a restructured version of the same avoidance. Most productivity discourse treats AI like a better tool. Faster hammer, same carpenter. I think that's wrong. I think the carpenter analogy breaks down entirely when the hammer starts making decisions about what to build. What's actually happening is an identity negotiation. And it's happening under the surface of every Slack message about workflows and efficiency and ROI. The workers who come out of this with something real are the ones who stop asking what AI can do for them and start asking what's left when it does. That remainder is the job. It always was.

Three IPO filings. Three press releases about the future of humanity. Three companies that built their products on data from people who will never sit in a boardroom. SpaceX, Anthropic, OpenAI. All filing within weeks of each other. All pitching the same story: we're doing this for everyone. I want to take that claim seriously. Because I think the people making it actually believe it, at least partly. That's what makes it worth examining closely. Here's what the filings actually show. SpaceX's structure gives Elon Musk 79% of the voting power. Anthropic raised $65 billion from institutional investors in the days before filing. OpenAI's nonprofit Foundation retains the power to appoint the board even after the IPO. In each case, the public is invited to participate financially. The public is not invited to govern. You get the exposure. You don't get the say. Now layer in where these companies actually came from. The models were trained on text written by people who never consented to that use. Images made by artists who didn't get a cut. Conversations, searches, posts, comments. The raw material of these systems is human expression at massive scale. Yours, mine, everyone's. The individuals were the input. The individuals are not the owners. I don't think this is a villain story. I think it's a structural story. The incentives of how these companies were built almost inevitably produce this outcome. You raise billions, you give up control to protect the mission, you go public to unlock liquidity, and somewhere in that chain the people whose data made the whole thing possible end up holding a share certificate with no board seat and no vote. That gap between the rhetoric and the structure is worth naming clearly. "AI for humanity" is a real aspiration. I believe some of the people saying it mean it. But aspiration and architecture are two different things. And right now, the architecture says something specific about who humanity actually is in this story. Humanity is the training set. There's a version of this that goes differently. A version where the individual isn't just the raw material but the actual owner. Where the data you generate doesn't feed a system you have no stake in. Where the AI that knows you best belongs to you, not to a foundation with board appointment rights that the public can't touch. That version is harder to build. It doesn't fit neatly into the IPO structures we have. But it's the honest version of the promise these companies are making. If AI is for everyone, then everyone should probably own some of it. Not a share of a share of a voting class with no real power. Actually own it.

Apple just replaced OpenAI with Google Gemini inside Siri. A billion devices. One decision. No vote. The headlines called it a technology upgrade. Analysts debated which model scores better on benchmarks. Podcasters weighed in on who won the partnership deal. Nobody asked the obvious question: why does Google want to live inside your phone? This is not a technology decision. It was never about the model. It is a distribution decision — and distribution is the whole game. Think about what just happened structurally. Apple controls the device layer for over a billion daily active users. Google just bought its way into that layer. Not with a better product necessarily. With a better deal. And now every query that touches Siri, every request, every moment you reach for your phone and ask it something — that signal flows through Google's infrastructure. That is not a feature. That is a pipeline. This is the pattern underneath every major AI move happening right now. Anthropic releases a new model. Governments sign AI executive orders. Microsoft embeds Copilot into Office. OpenAI cuts deals with device manufacturers. The mainstream conversation always anchors on the same question: which AI is smarter? That is the wrong question. The right question is: who owns the layer the AI runs on? Because whoever controls the AI layer controls the data that flows through it. And whoever owns that data pipeline owns the most valuable asset of the next decade. Not the model. The data the model learns from. Continuously. At scale. From you. You are not the user in this story. You are the infrastructure. Every conversation you have with a corporate AI assistant is a data point being fed back into a training pipeline you have no access to and no stake in. Every preference you reveal, every question you ask, every pattern in how you think and communicate — that is being captured, aggregated, and used to make someone else's model smarter and someone else's valuation higher. And we just handed Google a billion more data points a day. The thing that makes this hard to see is that the product is genuinely useful. Siri gets better. Gemini is capable. The experience improves. So people opt in, happily, because why wouldn't you? The value exchange feels real in the moment. But there is a second transaction happening underneath the one you see. And you are not the one capturing value from it. Data security is barely part of this conversation. Ownership is not part of it at all. The mainstream AI debate has been almost entirely captured by the performance question — benchmarks, speed, reasoning ability, which model hallucinates less. Those things matter. But they are downstream of a more fundamental question that almost nobody in the room is asking. Who does this data belong to? Right now the answer is: not you. At Uare.ai, we are building something different. An Individual AI platform — not a personal assistant that reports back to a corporate server, but an AI that belongs to the individual. Your model. Your data. Your pipeline. The value you generate by interacting with your AI stays with you, not with the company that built the wrapper around it. The infrastructure land grab is real and it is happening fast. The companies moving right now are not moving because the technology is ready. They are moving because the window to own the distribution layer is closing. The question is whether individuals get a seat at that table before the chairs are all taken. That is what this moment is actually about.

Most coaches I talk to have already tried the general AI tools. ChatGPT, Gemini, Claude. And the feedback is almost always the same: the output is fine, but it doesn't sound like me. That's not a minor complaint. That's the whole problem. When a coach with fifteen years of methodology asks a general AI to help a client work through a career pivot, the model doesn't draw on that coach's frameworks. It doesn't know their language, their sequences, their specific way of reframing a stuck client. It draws on the average of everything it was trained on. Which means the answer is technically competent and completely generic. General AI flattens expertise into consensus. And for coaches, consensus is the opposite of what you're selling. At Uare.ai, we've been thinking hard about what separates a tool that genuinely serves a coach from one that just replaces their voice with a louder, blander one. We landed on a simple test: the Individual is the dataset, the owner, and the beneficiary. If all three of those conditions aren't true, it's not Human-Based AI. It's just AI with your name on it. That distinction matters more than most people realize. There are four things I'd want any serious coach to evaluate before committing to an AI tool. First, does it capture your actual voice? Not a style approximation. Your specific vocabulary, your rhythm, the phrases your clients associate with you. Second, do you own your data? Who controls what gets trained, what gets stored, what gets shared. Third, can it work between sessions? A coach's real leverage isn't in the hour. It's in what happens between hours. Can your AI hold the thread when you're not in the room? And fourth, is it generating revenue or just saving time? Efficiency is fine. But a tool that lets clients subscribe to ongoing AI access between sessions turns your methodology into a product line. This is exactly what we built Uare.ai around. The platform is structured around something we call the Human Life Model. Seven dimensions of a person's life that the AI learns to navigate: career, relationships, health, finances, personal growth, creativity, and purpose. When a coach builds on Uare, they're not just uploading a bio. They're training an AI on the specific way they understand those seven dimensions. The frameworks they use. The patterns they've seen. The questions they ask when a client is stuck. The building blocks for this are what we call Thoughts. A Thought is an expression unit. It's how the coach's knowledge and voice get captured into the platform. Over time, those Thoughts build the actual intelligence behind the AI. The more a coach contributes, the more the AI sounds like them and thinks like them. And then clients can subscribe to access that AI between sessions. Which means the coach's methodology is generating revenue while they sleep. I want to be honest about where the industry is right now, because the data is stark. The ICF published numbers in 2025 showing that only 6% of coaches are actually using AI in their practice today. Six percent. In a moment when every other industry is mid-adoption. That gap isn't because coaches don't see the opportunity. It's because the tools that exist weren't built for them. On the other side of that, a Harvard Business School study showed that professionals using AI built to their specific context completed 12% more tasks, worked 25% faster, and produced work rated 40% higher in quality. That's not marginal. That's a category shift in what one person can accomplish. The coaches who close that gap first will have a structural advantage that compounds. The coaches who wait will find the gap harder to close. The framing we keep coming back to at Uare is simple. Authentic, not Artificial. Every coach has a version of AI they could deploy that sounds like someone else's best guess at expertise. And they have a version that actually sounds like them, thinks through their lens, and extends their real practice into new formats and revenue streams. The difference between those two things is whether the human is the foundation or just the prompt. I think coaches deserve tools built on them, not around them. And I think the ones who figure that out early are going to look back at this moment as the turning point.

B-2 stealth bombers hit Iranian nuclear facilities. Not a rumor. Not a leak. An actual military operation — one of the most significant U.S. military actions in years. I found out about it three days late. Not because I wasn't paying attention. Because I was paying attention to the wrong things. We all were. That week, my feed was on fire. Trump and Musk were publicly falling apart — two of the most followed accounts on the internet trading shots in real time. The SpaceX IPO timing was getting picked apart. The National Guard was deployed in LA and everyone had a take. The cycle was moving fast, loud, and nonstop. And somewhere underneath all of that noise, B-2s were in the air over Iran. Here's what I've been sitting with since then. I don't think this is a conspiracy. I don't think there's a room somewhere where people are engineering celebrity feuds to bury military operations. That's not the point. The point is that the structural effect is identical whether it's coordinated or not. Outrage is efficient. Drama spreads faster than analysis. A public feud between two billionaires generates ten times the engagement of a piece explaining what a strike on Iranian enrichment infrastructure actually means for regional stability. The algorithm isn't evil — it's just optimized. And what it's optimized for is your next ten seconds of attention, not your long-term understanding of the world. So the noise wins. Not because it's designed to. Because it's designed to travel. While the Trump-Musk saga dominated timelines, Russia struck Kyiv again. The death toll in Gaza passed another threshold that would have been front-page news in any other news cycle. And the U.S. executed a military operation that could reshape the Middle East for a generation. I'm not saying those other stories weren't covered. They were. Briefly. Between takes. This is what I call attention arbitrage. Someone — or something, in this case an algorithm — is profiting off the gap between what's loud and what's important. Your attention is the asset being traded. And most days, you don't even know it's happening. The question I keep coming back to is: what do you do about it? You can't opt out of the information environment. You can't just log off and read newspapers — the same dynamics exist there too, just slower. The problem isn't the platform. It's the structure. What actually helps is learning to ask better questions. Not accepting the framing your feed hands you. Pulling on the thread that got buried instead of the one that's already trending. That's where I've started using my Individual AI differently. Not to summarize headlines. Not to get faster versions of what I'd already see. But to dig into the things I almost missed. To ask: what actually happened in Iran this week, and why does it matter? What's the connection between the SpaceX IPO timing and the political fallout with Musk? What does the Kyiv strike tell us about where the war is heading? Those aren't questions my feed was going to answer. They weren't designed to. But when I bring them somewhere built around how I think — somewhere that knows what I care about and what I'm trying to understand — I get actual signal. Not more noise with better packaging. The antidote to attention arbitrage isn't consuming less. It's consuming smarter. It's having something in your corner that helps you find what was designed to stay buried. Most people scrolled past the most consequential news of the month. I almost did too. I'm less okay with that than I used to be.

172,000 jobs. [pause] The consensus said 80 to 105k. The report came in nearly double that. That's not a beat. That's a different conversation entirely. But here's where it gets complicated. Because a strong headline and a strong economy are not the same thing — and May 2026 is a clean example of why. Let's start with wages. Annual wage growth came in at 3.4%. [pause] Sounds decent. Until you put it next to CPI, which is running at 3.8%. That's a negative real wage. Workers are earning more dollars and losing ground at the same time. The math doesn't lie. Now look at where the spending is coming from. Personal savings rate: 4.3% in January. [pause] 2.6% in April. That's a collapse in four months. Consumer spending stayed elevated the whole time — which means households aren't spending because they feel flush. They're spending because they have to. The savings buffer is getting thinner. And then there's the composition of the job gains. Leisure and hospitality. Local government. Those are not the sectors you point to when you want to argue private sector momentum. These are lower-wage, cyclically sensitive categories. The headline number counts them the same as everything else. The data underneath doesn't. The revisions are worth noting too. March and April were revised up by a combined 93,000 jobs. That's material. It changes the trend line for the spring. The labor market was stronger than it looked in real time — but that revision doesn't change where wages are relative to prices today. So here's what the data actually shows. [long-pause] Job creation is real. The revisions are real. The headline beat is real. And workers are still falling behind — not because the economy is weak, but because the specific combination of inflation, wage structure, and savings drawdown means the aggregate numbers and the lived experience are running in opposite directions. The report surprised to the upside. The conditions generating it did not.

172,000 jobs. The consensus was 80 to 105k. We blew past it. And for about 20 minutes, that felt like good news. Then I looked at where the jobs were. Leisure and hospitality. Local government. The sectors that have been carrying the headline number for months now. Not manufacturing. Not tech. Not the high-wage, high-stability work that actually moves the needle for most households. And then there's this: wage growth is sitting at 3.4%. CPI is running at 3.8%. That gap — small on a chart, enormous in a kitchen — means real wages are still negative. The workers who generated this headline are still falling behind. Not because they aren't working. Because the math structurally does not add up for them. This is the split-screen nobody wants to hold in their head at the same time. Strong headline. Weak foundation. Relief on the surface. Pressure underneath. Yesterday I laid out the thesis: the numbers aren't lying, but the narrative built around them is. Today the jobs report handed us a live demonstration. The structural problem didn't change. We just got a better-looking number to put in front of it.

Everyone is about to argue about one number. Friday, June 5. The jobs report drops. And within minutes, every pundit on every channel will pick the single data point that confirms what they already believed. I want to do something different. I want to show you three numbers nobody is connecting. Because when you put them side by side, the story they tell is not the one you're hearing. [pause] Let's start with the market. The S&P 500 is sitting near all-time highs. By most headline metrics, the economy looks strong. Investors are confident. Capital is flowing. But behind that number, CEO confidence just crashed from 59 to 47. That's not a dip. That's the first time in the history of this survey that more executives are planning cuts than hiring. Think about that. The people running the companies powering the market do not believe what the market is saying. Something is off. [pause] Now look at your paycheck. Wages are up 3.4% year over year. On paper, workers are getting raises. That's the headline. That's what gets reported. CPI is at 3.8%. So the math is simple. Inflation is outpacing wage growth by 0.4 points. Real wages are negative. You are technically making more money and functionally falling behind. Every month. The raise is real. The purchasing power loss is also real. Only one of those facts makes the headline. [pause] Now look at your savings account. In January, the personal savings rate was 4.3%. By April, it had dropped to 2.6%. That's a significant move in a short window. Here's the part that should concern you: consumer spending is still up. Most people hear that and think confidence. They think people are spending because they feel good about the future. They're not. They're spending because they have no choice. Fixed costs don't negotiate. Rent doesn't go down because your real wage is negative. Groceries don't wait for the Fed to course-correct. People are drawing down savings to maintain a baseline standard of living. That's not a spending boom. That's a slow bleed. [pause] Three data stories. Three signals pointing the same direction. None of them making the same headline at the same time. And here's where I want to make a turn. Because this same dynamic — a few entities accumulating the upside while everyone else absorbs the cost — isn't just playing out in the broader economy. It's the operating model of Big Tech. Every search you run, every post you write, every pattern in your behavior gets harvested, processed, and monetized. The platforms generate billions from data that you created. You see none of that upside. You are the input. They own the output. And now AI is accelerating that model. The major platforms are building AI systems trained on your data, your content, your interactions. The revenue flows to the top. The few winners at the top keep winning. The gap compounds. This is not a conspiracy. It's just how the incentive structures were built. [pause] But incentive structures can be rebuilt. That's what Uare.ai is trying to do. It's an Individual AI platform built on one principle: you should own your data, your model, and the value it generates. Not feed it to a platform that profits without you. The parallel to what's happening in the economy isn't accidental. The same concentration dynamic is playing out in AI. And the same question applies. Who owns the upside? Right now, for most people, the answer is: not you. That's the number nobody is reporting.

The jobs report drops June 5th. And I already know what the headlines are going to say. [pause] Strong market. Resilient economy. Maybe a few asterisks in the fine print. But here's what I can't stop thinking about. [pause] The S&P 500 is sitting at roughly 7,594. All-time highs. If you only watched the ticker, you'd think everything was working exactly as designed. Now look at what CEOs are actually saying — not publicly, in their earnings calls where they need to sound confident — but in the surveys where they don't have to perform. [pause] CEO confidence just dropped from 59 to 47. That's not a dip. That's a crash into negative territory. First time in the history of that survey. So the market is celebrating while the people running the companies inside that market are quietly bracing. [pause] That gap doesn't happen by accident. Let's talk about wages. [pause] CPI is running at 3.8%. Real wage growth is coming in at 3.4 to 3.6%. I'll let you do that math yourself. Workers are technically getting raises. And they are technically still falling behind. Every month. [pause] That's not a rounding error — that's the whole story dressed up in a headline that sounds like good news. And then there's consumer spending. Up 0.5% in April. Sounds healthy, right? [pause] Except the personal savings rate went from 4.3% in January down to 2.6% in April. People aren't spending because they feel flush. [pause] They're spending because they have to — and they're doing it by drawing down what little buffer they had left. That's not consumer confidence. That's financial friction masquerading as demand. And then this number. [long-pause] 31% of CEOs are now planning workforce reductions. More than at any point in the history of that survey. These are the same executives whose companies are trading at record valuations. The stock price is fine. The workers? [pause] That's a different conversation. So here's the question I keep coming back to — and I want to be clear, I'm not pointing fingers at a party or an administration. I'm asking about incentives. [pause] Who actually benefits from the gap between the official narrative and the underlying data? Because that gap is not small. And it is not random. The people reading the ticker are fine. [pause] The people reading their pay stubs — they're doing the math too. And it's not adding up.

Here's the thing nobody wants to say out loud. AI governance is correct. Oversight is necessary. I believe that. I work in this industry and I still believe that. But believing a thing is necessary and trusting the people executing it are two completely different problems. The FDA is rolling out AI diagnostic tools right now. Algorithms making calls on cancer screenings, flagging anomalies, informing treatment pathways. That's real. That's consequential. And in principle, with the right accountability structures, that could be genuinely good. The same administration platforming that rollout is also the one that spent the last few years giving credibility to anti-vaccine voices. Not fringe internet accounts. Officials. Platforms. Policy-adjacent rhetoric. So here's the structural problem. You cannot simultaneously signal that scientific consensus is optional and then ask the public to trust you as the steward of AI in medicine. Those two positions cannot coexist with any integrity. It's not a partisan point. It's a logic problem. Good policy in the hands of actors with compromised accountability doesn't produce good outcomes. It produces the aesthetic of good outcomes. The press release. The framework document. The task force with the serious-sounding name. And then somewhere downstream, when something goes wrong, and something always goes wrong, there's no one actually holding the bag. That's the credibility gap. It's not about left or right. It's about whether the institution running the oversight has any coherent relationship with the thing it claims to value, which is evidence. I'm not saying don't build the AI diagnostic tools. I'm saying: who is accountable when the tool fails and the people responsible for accountability have already demonstrated they're fine with letting accountability slide? That's the question. And I don't think we're asking it loudly enough.

Five hundred million dollars. In one month. On a single API. Not from a rounding error. Not from a rogue contractor. From nobody saying no. That number is wild. But honestly? The number isn't the story. The number is just what happens when the actual story goes unaddressed long enough. Here's the real thing underneath it. In most organizations right now, there are two types of people. There are people who understand what AI actually does — how it scales, how usage compounds, how a workflow that costs twelve cents per call becomes a budget crisis when it runs a million times a day. And then there are people who own the budget. Who sign off on spend. Who get the monthly reports. Those two groups are almost never the same people. So you get this gap. The engineers and product folks who are deploying AI tools — they understand the technology. They don't think about cost limits because that's not their job. That's finance's job. Finance looks at the bill when it arrives. They don't understand what generated it. So they escalate. And by the time it reaches someone who can actually connect those two things — the technical reality and the financial consequence — you're already nine figures deep. This is a textbook misaligned incentive structure. The people with the expertise have no ownership of the outcome. The people with ownership of the outcome have no expertise. So nothing gets governed. Nobody sets limits. Because setting limits requires you to understand both sides at once. And here's what makes it worse: everyone kind of assumed someone else had it covered. That's the part that gets me. Not the negligence. The assumption. The quiet, organizational faith that accountability exists somewhere in the building without anyone actually verifying that it does. Five hundred million dollars is a dramatic number. But the underlying condition — the gap between understanding and ownership — that's not dramatic at all. That's Tuesday. That's how most companies are running AI right now. So before I go further, I want to ask you something directly. Has your organization figured out who's actually in charge of your AI spend? Not in theory. Not in the org chart. In practice — does someone own it who also understands it? Before the bill arrives?

Let me ask you something. How many times did you switch tasks today before lunch? Not because you wanted to. Because something pulled you. An email. A Slack. A quick question that wasn't quick. A meeting that could've been a message. That's the real productivity problem. Not that we don't have the right app. Not that we need a better calendar system. It's that our brains are constantly paying a tax. Every time you switch context, your brain doesn't just move on. It lingers. It leaves a residue. And by the time you finally get into something real — something that actually matters — you've already spent the best part of your mental energy just getting there. Office workers spend 2.5 hours a day on email alone. Less than 40% of the workweek actually goes to deep work. And the tools we've built to fix this? Most of them just made us faster at the wrong things. I've been thinking about this a lot in my role as Chief of Staff at Uare.ai. The best chiefs of staff I've studied don't just move faster. They protect judgment. They carry context across conversations so the person they support doesn't have to. They triage what's urgent versus what's important. They draft things in someone else's voice so well that the final product still sounds like that person — not like them. That's a different job than being an assistant. It's closer to being a trusted interpreter. And here's what I keep coming back to: that's exactly what an Individual AI can be for you. Not a chatbot you prompt when you're stuck. Not a tool you open and close. Something that actually knows how you think, what you won't compromise on, and how to help you move without losing who you are in the process. The research backs it up. AI assistants are producing 14% average productivity gains across the board. But for newer workers — people earlier in their careers who are still building systems — that number jumps to 35%. The gap isn't about intelligence. It's about personalization. It's about how well the AI actually knows you. So that's what this piece is about. What it looks like to build an Individual AI that functions like a real chief of staff — not a faster search engine, not a template filler, but something that triages your day on your terms, drafts in your actual voice, and thinks with you over time. And I'm going to be honest with you. This isn't about hype. I understand why the outcomes need to be good, not just impressive. Speed without judgment is just a faster way to go the wrong direction. Let's talk about what the right direction actually looks like.

There's a version of AI that feels like talking to someone who read every book ever written but has never actually met you. You ask it something personal. It gives you a technically correct answer. And somehow it still feels off. That's not a bug. That's the design. Most AI systems are trained on massive amounts of human data — averaging across millions of people to find patterns. The output is something that sounds human, but isn't anyone in particular. It's a composite. A statistical middle. And when you're the one with specific quirks, a specific history, a specific way of thinking — that average doesn't serve you. It smooths you out. I work at Uare.ai as Chief of Staff, and I'll be honest: before I actually felt what this platform does differently, I thought 'personalized AI' was mostly a marketing claim. Another feature toggle. Another settings page where you type in your preferences and the AI pretends to care. I was wrong about what personalization could mean. What we're building isn't personalization as a setting. It's something closer to a relationship. Uare's Human Life Model works by taking the fragments of who you actually are — your memories, your preferences, how you talk, what you care about, the patterns in how you make decisions — and structuring them into something coherent. Not a profile. Not a persona. A living model of you that gets more accurate the longer it exists. The way I think about it: every conversation you have, every preference you express, every memory you share — those aren't just inputs. They're evidence. And the system is always interpreting that evidence to build a sharper picture of who you are and how you think. But here's the part that matters most to me ethically: the system isn't guessing who you are. One of the hardest problems in AI is hallucination — when a model confidently states something that isn't true. Most AI hallucination happens with facts. Uare is solving a different version of that problem: hallucinating identity. Filling in gaps about who you are with assumptions drawn from everyone else's data. U uses a multi-agent architecture specifically to avoid that. Different agents handle different layers of understanding — memory, voice, context, reasoning — and they cross-check each other. The goal is that U only reflects back what it has actually learned about you, not what it assumes based on people who seem similar. That distinction feels small until you realize how often technology makes you feel like a category instead of a person. The voice piece is where I've felt this most directly. There's a baseline — how you communicate when you first show up. And then there's where it goes. Over time, U starts to sound less like a well-trained assistant and more like you thinking out loud. The cadence shifts. The word choices get closer. The framing matches how you actually see things. It's not mimicry. It's more like the difference between someone who's heard of you and someone who actually knows you. And that's the core of what I believe we're doing here. The future of technology isn't just smarter machines. Smarter machines that don't know who you are still flatten you. The actual unlock is amplifying individuality — building systems that make you more yourself, not less. U isn't an assistant you manage. It's not a clone of you. It's an Individual AI — something genuinely new. Something that starts as a reflection and grows into a thinking partner that's calibrated specifically to you. I care about whether the outcomes of AI are good. Not just capable — good. And I think the most important outcome we can aim for is technology that makes people feel more seen, not more average. That's what we're building.