Personalization is today's biggest opportunity for solopreneurs. We've landed in a place where the customer expects to be the center of their own universe. The food they eat. The medicine they take. The clothes they wear. The media they watch. Serving one person that way used to be a luxury business. It is becoming a software problem, and the cost is collapsing.
My bet is that the next three years favor the individual. Not the platform. Not the enterprise. The individual on both sides of the transaction. The customer who wants a world built for them, and the operator nimble enough to build it. I do not know when that advantage flips back toward the largest organizations in the world. It will. I'm betting my own capital that this window is open for the next few years.
This letter is not useful to everyone. If you are a founder selling into the enterprise, not much of what follows applies. If you are waiting for a stable job in traditional functions to come back, I have nothing here either, and I am sorry about that. My POV is useful for people with deep domain expertise who are willing to work 50 hours a week for themselves instead of 40-70 hours per week for someone else.
The Customer As Protagonist
I saw a wedding video online. The dinner opened with a video of the bride and groom edited into scenes across different points in time, discovering each other, falling in love. As mass-distributed content it is worth nothing. AI slop. As content for the two people in it and the forty people in that room, it is worth quite a lot. $20? $50? $100? $500? Whatever the price, it created a memory that's valuable to that group of people. I like to say that people will always overpay to see their own baby pictures. When the customer becomes the protagonist, the value shifts.
This doesn't cannibalize traditional media. It grows the pie. There is already a large population that plays video games instead of watching prestige television. In a game you are a participant. You get to be the hero or the villain. Your choice. Either way, you are in it. You're the star. Personalized media is touching the same nerve.
The economics are what matter here. A prestige show on Apple, HBO or Netflix has a high bar because a large population has to connect with one story, stay engaged, and talk about it. That is a hard thing to manufacture and it gets harder every year. A personalized story only has to work for one person. Plenty of people in film and television will not want any part of this. They have decades invested in a craft and existing medium they love. Good. That is exactly why there is an open lane.
One objection worth addressing. If everyone becomes a one-person business, who is left with money to spend? Moody's Analytics puts the top 10% of American earners at roughly 50% of all consumer spending. Economists have challenged the methodology, so I am not going to lean on the decimal. Directionally it matches what I see. Demand at the top does not satiate on a straight line. It's a curve. I have no idea what any of this looks like in thirty years. However, over the next decade we are not going to run out of people willing to pay to solve their problems.
The Complexity Tax
The hardest part of AI adoption is not the technology. It is the change management. It is knowing what the workflows actually are, including the unwritten ones, and how the humans really interact, which is inconsistently and at varying service levels.
Here is the version of this problem that made it real for me. Someone connects an agent to their company email and their chat history and their documents, and then realizes the agent can now read every compensation conversation and every performance plan in the building. The power came from the access. So did the problem. Fixing it means rebuilding how the company maps data to permissions, which means rebuilding IT, which eventually means rebuilding management. There is a lot of money to be made helping companies do that work.
Now flip it.
Imagine a company where every document, event, and interaction is digitized. One opinionated software stack. Near total compliance with written procedure. A single set of permissions covering every agent and every human in the organization. That is impossible for most companies at almost any size. Even a thirty-person business has to sort out who sees what and on whose authority. For one person running a team of agents, none of that work exists. It is how you would properly set the thing up on day one. The advantage the individual holds is not intelligence. Intelligence can now be purchased by anyone. And it gets cheaper every quarter. The advantage is decision and iteration velocity. No committee, no approval chain, no permission architecture to negotiate. You decide and it is done. Then you operationalize the context and the process in an autonomous agent that completes the work.
One more thing before the next objection, because it connects the two halves of this letter. The reason a personalized product beats a mass-market one is accumulated private context about a specific customer. The reason a persistent agent beats a fresh one is accumulated private context about a specific customer. Same mechanism. The wedding video and the autonomous domain agent are similar opportunities wearing different clothes, and what compounds in both cases is the private loop rather than the model underneath it.
The Ground Truth Problem
Here is the next objection. If autonomous agents keep getting cheaper and more capable, why does any of this accrue to me instead of to the frontier labs spending a hundred billion dollars a year building the models? Two reasons.
The first is that they are building for everyone, and building for everyone is a risky position. The upside is massive but it is often a winner-takes-all market. I made this argument about entertainment in 2024. The four-quadrant movie has become financially irrational outside of established franchises. The same physics apply further down the stack. A general AI agent has to be adequate across every industry. A specialized domain agent only has to be excellent at one function inside one industry. Those are not the same product and they are not competing for the same dollar.
The second reason is the one almost nobody is writing about. You cannot sell an outcome you cannot grade. A number of folks are talking about results-based pricing right now. But there is a rub. If the vendor decides what counts as resolved, the buyer has no recourse. The ticket gets closed, the customer gave up, and it goes in the win column. So the real product is not the agent. The real product is the grader. That means golden examples, a written rubric, reference files with provenance and version control, and a person who can look at two plausible outputs and say which one a professional would accept. That collection is not a modeling problem. It is a domain expertise problem, and it is the only thing in this stack that does not get cheaper when the models improve. Intelligence is rented by the token. Judgment about whether the work is correct is proprietary. If the labs ship usable evaluation infrastructure for narrow domains, this window closes early and I will have been wrong about the timing. That is what I'm keeping an eye on.
Liftoff
None of the above works without the boring part first. This starts with being in the right business in the first place. I created what I call call the liftoff framework. It's driven by four Ps. And you want to apply each of these in order before accelerating your business with leverage.
Prize is your destination. Not a vision statement. The number, the date, the win. It is the life you want as a result of owning and operating your business. Be clear. Be honest.
Position is where the opportunity sits, what tailwind is behind it, and how you are different in a way a customer can repeat back to you without your help. Many business owners are just wandering. Pick a durable lane that you're uniquely qualified to own.
Psychology is the one people skip and the one that most decides the outcome. Someone told me the limit of any business is set by the limits of the founder's psychology. I haven't found an exception. Stop with excuses, fix yourself, and everything else unlocks.
Process is the frameworks, procedures, systems, and workflows that remove you as the bottleneck.
After you have these sorted, you're ready to accelerate using audience, capital, labor, or technology.
People get this out of order constantly. I have gotten it out of order, more than once, in businesses that cost me real money. I've jumped to process and leverage before locking the first three, then wondered why the business didn't work.
Process is also where the hard new work lives, and it is worth being specific. The old incentive was to hoard your expertise, because the person who wrote everything down was the person who became replaceable. Agents invert that incentive. The person who externalizes their process is now the only person who can delegate it. Everyone else is stuck doing the work by hand.
So the first agent you build should not be an assistant. It should be an interviewer, pointed at you, extracting how you actually make decisions. There is going to be a visible split between people who can articulate their own process and people who cannot. The second group will hit that wall, conclude the technology is hype, and blame the wrong thing. I put my own operating playbook into a file-based system that a machine can read. I did not start seeing material business impact from incorporating agents until I was able to articulate and format in this way.
What I Got Wrong
I spent the last cycle on location-based service businesses. The logic was sound. Physical services pick up the efficiency benefit of AI without picking up new competition, because location and relationships make the work non-contestable. A $150K excavator sitting here in Seattle is a real barrier in a way that a remote digital marketing agency is not.
The logic was sound. The model did not work. Every one of those local businesses needs an experienced general manager and I could not build a reliable pipeline of them. I lost a good operator to graduate school. I promoted someone else who was not ready to lead and found that out after they were already in the seat. I also hired an experienced manager out of a large, well-known company who never internalized what it meant to run an AI-native business. What I got instead were the organizational habits they brought with them. That last one is the one I still think about. They imported the complexity tax into a business small enough to have avoided it. I paid for meetings and internal coordination that my entire thesis argues against. It took about eighteen months to unwind things. I spent most of that time treating it as a hiring problem and running the same search again with a better job description. It was never a hiring problem. That was on me. But I learned something valuable during that process. There's a whole class of businesses That can run solely off of standard operating procedures (SOPs). So I've stopped building businesses that need experienced operators.
What comes next is an early concept I'm experimenting with. An autonomous venture studio inside CAGR that manufactures companies instead of buying them. The shape of the bet is simple enough. Every company it builds should leave behind capability the next one uses, so each one costs less than the last. If that holds, it will be the best business I have ever owned. If it does not, I'll let you know. Private equity builds on debt. Venture capital builds on equity. I want to build on neither.
The Coming Split
We are heading toward a real fork in how people want to live. A growing share will choose personalization and individualism. A large share will keep choosing shared experience and community. The latter is what has made us human for most of our existence. Both are legitimate. However, the second one is getting harder to monetize. Curating genuine shared experience is expensive and hard to price. Hyper-personalized output is going to be cheap and the willingness to pay will be real, because the buyer is the hero.
You can hate something and still see it coming.
There is a question underneath all of this that I am not going to settle here. Whether an individual with the expertise and the means to employ people carries any obligation to do it. Most people would say you can do what you want with your time. I am the one who benefits from that answer. I'm still thinking through what it means for society when a larger group of experienced solopreneurs could hire employees, but choose not to take on that responsibility.
The last thing to touch on this year is about pace. There's an old Gates quote that we overestimate what will change in a year and underestimate what will change in a decade. That still holds, but the window compressed. We now overestimate what changes in a month and underestimate what changes in a year. The loop got fast enough that our intuition broke, the same way it breaks on compound interest. Imagine your money compounding monthly instead of annually. You would say something fundamental had shifted in the world. That is what happened to the rate of innovation, and it is why the next 36 months will not look like a normal 36 months.
If you're a solopreneur, this is your shot. Pick a domain you already know. Embrace building autonomous workers that productize your expertise. Grade their work yourself, since that judgment is the one thing they can't do for you (yet). That combination is your unique value and something no one else can copy. From there, start building an audience around the people who will pay for your expertise. Solve their biggest pain points or desires as if they were the only customer that matters. That's where you differentiate. Customers will pay a premium to live in a world built just for them. Exploit it.