How I'd build a one-person business if I started over in 2026

Info products are dead & AI changes the game

Watch the video version of this letter on YouTube.

Listen to the audio version of this letter on Spotify.

---

Everything is changing.

Business, the internet, social media, coding, art, everything. What worked just last year is much less likely to work this year.

I’m not going to tell you to start an agency. I’m not going to tell you to start freelancing or coaching or any of the other low-hanging fruit that I’ve already talked about plenty of times in depth.

Instead, I’m going to show you exactly what type of business makes sense right now.

I’ve been seeing people say that “value-based content is dead” and that “info products are dead.” People feel like it isn’t worth it to start as a creator when there’s a new paradigm emerging that they don’t understand yet (I don’t think anyone truly understands it. I don’t. This letter is an educated guess at best).

Let’s talk about how the one-person business is changing, what the future of education looks like, and how to actually take advantage of this.

I – The one-person business model is evolving

Fortunes require leverage. Business leverage comes from capital, people, and products with no marginal cost of replication (code and media).

– Naval

For the last decade or so, the best way to build a cash-flow, absurdly high-margin business was with information products or coaching services.

It was great because with the internet, social media, and software, one singular person could start an entire business. That’s insane. One SINGLE person could be their own marketing, sales, product, and design departments. And this is obviously more possible than it ever has been with agentic AI - as we’ll continue to dive into in future letters.

In this digital world, you write content to get traffic without needing a physical business in the right location. You do something interesting with your life, or learn one of your passions enough until you become a valuable resource, and then you share it with others in an educational or entertaining way. Your brand is you, your content are little pieces of your mind, and your products are processes that solved a problem in your life and led to something better. Many overcomplicate that. This isn’t something that will ever go away, but the vessel by which you do this will continue to change, so you will need to keep your finger on the pulse of how everything evolves.

My entire goal with these letters is to help you become future-proof. They’re built around 2 pillars:

1. How do you figure out what you want in life (psychology, philosophy, personal development)

2. How do you succeed with that in today’s world to reach that (skill acquisition, business, technology, AI)

Previously, the second part was largely handled with a personal brand on social media and with some kind of information product, be it a simple ebook (level 1), a course (level 2), a cohort (level 3), a community (level 4).

But since everyone is talking about how “info products and value-based content are dead,” I think there’s something deeper at play. The process is still the same. You still pursue your curiosities and interests and share that with other people, but the vessel for doing that is no longer a static PDF, video course, or community.

I want to make this very clear: I am not on the jaded side that thinks every info product is a scam because everyone has one now. I think they can be life changing, they have been for me. I think they are the antidote to negative aspects of traditional education (i.e. conformity and getting trained into a job you hate). I do not think they are going away for good, and what I say in this letter will not apply to every single industry or niche, but as with almost every career sector right now (thanks to AI) the baseline that gets results has been raised.

An average person at a low level of development can now take the “quick and easy route” and tell AI to create a persuasive ebook, write viral content, and generate images and assets. They can make some form of money, but they will never compete with those who understand what’s going on.

In other words, it has never been easier to start a one-person business.

But if you don’t want to be in the lower class of the creator economy (I hate saying that, but it absolutely exists), you need to understand something very important.

II – We are breaching into stage 6 of the market

Markets move through predictable stages of awareness and saturation.

Eugene Schwartz categorized these into 5 stages.

At level one, when nobody is in the market yet, you just state what your product does and that tends to do the job. Then competitors arrive and you have to make bigger claims. Then the market gets skeptical so you explain your mechanism (saying “I will help you make money online” started to sound like a scam and you had to be more specific like “Implement the 2 Hour Content Ecosystem”). Then competitors copy that mechanism. Finally, everyone is exhausted by claims altogether and brand starts to be the major differentiator, which is happening right now with personal branding. People crave the sense of belonging they get from joining a tribe with a mission (I’ve talked about how your mission is your niche before). Community and authenticity matter more than who has the best sounding product.

We’ve hit that final stage across the board. So much so that “authenticity” isn’t a differentiating factor. People are just tired of the same old courses and coaching. It sucks, because those things are impactful, but there’s more to it.

The select few who are smart enough to transcend info products as a whole will be the ones who win. Transcending doesn’t mean you leave them behind. It means you evolve to the next level and integrate them, because education/learning as a domain will never go away.

But since you are competing in a space as one singular person, things can become saturated quite quickly. What I am going to tell you isn’t going to secure you for long, because the next level will be just around the corner. Info products didn’t last long (maybe 10-15 years?), and the next phase won’t last that long (maybe 2-3 years?). Who knows how business is going to look after that. We will be building for tomorrow rather than a year from now. Iteration cycles will be mind-bogglingly short. Timescales will be compressed.

The main advantage of being a one-person business is speed and adaptability. You must constantly be moving and iterating until you reach escape velocity and can move beyond the one-person business.

Of course, AI will make it so people can copy the front runners faster, but there is some good news here:

Most people just don’t do anything.

That sounds harsh, but you’re probably in an internet echo chamber. The average person still treats AI as a place to ask questions or generate a pretty image, and that may never change because the average person doesn’t use their free time to learn and explore. The average person thinks that they can do one thing, see some success, and then never do anything again because they are working toward an illusory retirement where all of their effort pays off and they can just “chill” for the rest of their life. The mechanical/linear model of work is not something to bet your future on.

Meaning is generated from struggle, status, and curiosity.

The day you stop following either one of those drives is the day your life gets considerably worse.

III – The future of education (products)

It’s no longer about sitting in front of a government-trained expert and ‘learning’ the same thing as everyone else to end up with a soon-to-be irrelevant skill stack... It’s about finding someone you relate with. Someone with a shared vision for the future.

– Purpose & Profit

I’ve always held a firm belief – even when everyone else considered courses and coaching a scam – that the creator economy was an interest-based education system far more effective than traditional schooling at teaching people the relevant skills to create the life they want.

I believe that people learn more, faster, and better when they learn from someone they can relate with. I believe that anyone can follow 4-5 people on a specific topic and become an expert in that domain much faster than they would by studying it in a formal setting.

I still hold the belief that there is knowledge and experience that only you can acquire, and I think it’s noble to both acquire it and pass it down. Some would even say that’s the most meaningful way to live, depending on which philosopher you look to.

But the format or vessel of how we pass down that knowledge needs to evolve as it always has from speaking by the fire to sending letters to building libraries to the day everything changed when the internet was born.

Static courses don’t cut it anymore (well they do, it’s not like they’re just going to zero one day, but the general market sentiment is a bit burnt out from them). A 10-hour video library where someone watches, takes notes, and hopefully implements isn’t something that people want that much anymore. It feels slow. They crave something faster and more efficient. Plus, it doesn’t help that 90% of people don’t finish courses and even more don’t get results. And now, with AI, anyone can generate that information in a second (if they know what to ask, which is unlikely).

That said, the future of education, in my eyes, is learning experiences rather than static courses.

You’re not only selling information anymore. You’re practically selling a second version of your mind. You’re selling your coaching services, but you aren’t there. Instead, you pass off everything you know to the AI and let people learn at their own speed.

“But what about schools who are implementing AI like this? Aren’t you competing with them?”

No, because schools are educating the public on what’s necessary to operate within society. Preferably, you’re teaching people something very few people have done and schools wouldn’t think to teach. You are placing a price tag on your own curiosity and self-development.

IV – How AI changes the way we learn

AI is great for utility-based tasks, but not so much with meaning-based tasks unless you find a way to implement it in your workflow in a useful way. (No, I don’t think AI makes everyone dumb and takes away all of our human faculties. I think people can absolutely fall into that trap on an individual level if they do not have any restraint or desire to maintain those faculties).

As an example, I had AI do a ton of writing for me last month. Around 30 articles in 2 hours. But this isn’t the typical writing (like this letter) where I would have to be so much more critical of what the AI generates.

For this little project, I built a help center so I could have our support agent reference the knowledge base of 30 articles and answer questions or resolve support requests quickly. This knowledge base is also used by our new AI agent that can quickly resolve questions that don’t need a human. Response times are now instant.

To build this help center, I told AI to research how other help centers are structured, propose ideas for articles that should be there, and interview me about my company and product to write all of them. All I did was answer questions and fix some mistakes that the AI made when talking about how to do specific things within the product, Eden.

Now take that same concept and apply it to the internet education/creator economy.

Imagine a course that has a knowledge base of all the course content, but the chatbot isn’t for support. It’s there to help the student learn, practice, and implement the material. Hell, it can even help them do the work.

An AI that helps you write the newsletter like a coach sitting next to you can get results faster than a student learning from a static course and potentially quitting because they got stuck on something.

This is what I mean by learning experiences. The student no longer just buys a bundle of lessons and prays that they change their life. Instead, they’re interacting, getting feedback, and actually doing the thing with guidance. Learning by doing is the best way to learn, as I’m sure you know. Gone are the days of being stuck in tutorial hell, hopefully.

This isn’t new, by the way.

Before mass education existed, this is exactly how knowledge was passed down. Apprenticeship. A blacksmith didn’t hand his apprentice a manual and say “figure it out.” He worked alongside him. Corrected his grip in real time. Pointed out mistakes as they happened.

Then we industrialized education. We needed to train thousands of workers quickly, so we created the lecture model. One teacher, many students, standardized curriculum. Efficient for scale. Terrible for actual learning (but turned out to be great for indoctrinating).

Courses are just digital lectures. And they have the same problem. The teacher isn’t there when the student gets stuck, and the education happens outside of the environment in which the education is to be applied.

For the first time since the apprenticeship model, we can have personalized, interactive guidance at scale. This won’t click for those who think that all AI output is the same as you chatting with ChatGPT. Your AI doesn’t replace you as the teacher. It’s yet another way to ‘build once, sell twice’ but on a more impactful scale.

If I were personally doing this (I don’t plan to because I’m building two bigger companies right now. I think the one-person business is an incredible starting point, but I want to keep pushing) here’s what I would do with my 2 Hour Writer course.

First, I would create the modules to the course. That’s already done on my end.

Then, I would build some form of an interactive chatbot. Like a small software. This would take me a bit because I’m not a programmer, but with today’s tools I don’t think something that simple would be too difficult to do.

It would have 3 different modes or tabs: Learn, Practice, Create. They would each have a series of prompts that can be executed in the right order to teach, give interactive practice, and “grade” their work so to say. It would make learning and writing very fun.

V – How to build a micro SaaS (non-exhaustive)

Again, I don’t think info products will be (completely) dead anytime soon.

Education is too important for the human brain, and choosing what to be educated on based on goals you derived for yourself is still one of the last moats. Learning is too foundational to the human experience to ever be fully commoditized.

With that said, the info products we know today will look more like software over the next few years. Rather than having an ebook to download, you have a website to visit or app to install. A few of my business friends are already doing this. One turned his offer creation process into an app that helps you find the right idea and formulate your offer.

So, rather than selling something like a “how to talk to girls” ebook, course, or coaching program, you create a chatbot that simulates them talking to girls. Rather than a static productivity course, you create an AI that helps them identify their vision, goals, and priority tasks, then pings them with notifications to actually do those things. Maybe integrate that with some kind of todo list app and now you have something much more enticing than “buy my course” to promote in your content. You build a coach that costs far less than it would to hire you.

For something like a high-ticket service rather than a product, this could look like helping other businesses implement an AI system in a unique way. You could consult people on a content writing process or build an agency as you normally would while offering what you would have offered before, but with AI integrated.

If you’ve been following along my newsletters at all, or if you’ve looked at the prompts I have on here, then you’ll know exactly what I’m getting at. Any of the prompts I’ve sent to this list could easily become a micro software that I charge a few bucks a month for. Or, if it’s a one-and-done type deal like my friend’s offer creation app, it could be a one-time price to access. I also taught how to create good prompts here (which will be a crucial step in this process). Those would become the system prompts of your AI portion of the software.

Rather than creating a very general ChatGPT, you create a hyper-specific chat or software that helps people learn, practice, and do the thing you were teaching about in the first place with the info product.

“But Dan, isn’t building a ChatGPT wrapper cheating?”

Well, by that logic, building anything on the internet is cheating. Typeform, a billion dollar company, is just an HTML wrapper. Cursor is a GPT wrapper. Any info product you build is wrapped by the platform you build it on, and if AI is like the next cloud solution, every piece of software on the internet will be a wrapper. So yes, as one person who has to be savvy with their resources and use third party tools, a ChatGPT wrapper is a great way to go. And they tend to be pretty simple.

People telling you that you can’t create a ChatGPT wrapper is like telling you that if you want to build a chair, you can’t use wood.

You create the course kind of like you would create a help center. You build out all of the articles – you practically create the course, with the help of AI if you want, but with a few more steps.

Then you create what’s called a system prompt.

Like we just talked about, a system prompt is the set of instructions you give to an AI before it talks to anyone. It defines the AI’s personality, knowledge boundaries, and behavior rules. Think of it as programming the AI’s identity.

For example, if you were creating an AI writing coach, your system prompt might include:

• The AI’s role: How the AI should act

• Its instructions: What exactly it should do, step by step

• Its knowledge base: All of your frameworks, processes, and examples

• Its boundaries: What it should and shouldn’t help with

• Its personality: How it talks, what tone it uses, how harsh or supportive it is

The prompt is what makes your AI different from anyone else’s. It’s your specific knowledge, your frameworks, and your voice packaged into a tool that can help thousands of people at once.

Can people steal it? Sure they can. But they could have stolen anything that you build. You have to act fast, as I discussed earlier.

VI – Obsession and experimentation are the moat

Use a tool like Replit or Cursor.

Replit is more beginner friendly, Cursor is a bit more complex.

You can also try playing around with just Claude Code and see how that does.

Yes, you will have to learn how to use it (shocking, I know, I can’t believe you have to actually spend time learning something to get a different results from those who don’t learn anything). The best way to do this is to build alongside something like a YouTube tutorial and use AI as a study partner when you get stumped. Ask it what the best way to go about “vibe coding” is. Ask it to create a comprehensive game plan to build the software step by step. Have it ask clarifying questions about how you want it built.

Then, just start tinkering and building.

Your first few tries will obviously suck and you will have to learn how to fix it. You will probably have to build 3-4 apps before you feel like one is worth putting a price tag on.

The question now is, where does your actual advantage lie?

Your advantage isn’t doing what AI can’t do. That’s a losing game.

Your advantage is doing what only you would think to do with AI.

Not everyone types the same thing into ChatGPT or Replit or Cursor. Meaning the output of those things are infinitely unique. Your unique combination of interests, experiences, and insights leads to prompts and products that nobody else would create.

Naval explains this as specific knowledge:

• Specific knowledge can’t be trained for

• It comes from pursuing genuine curiosity

• Building it will “feel like play to you”

• It’s often at the “edge of knowledge”

The person who spent 10 years obsessing over productivity and then builds an AI tool around that specific knowledge will create something fundamentally different than someone who just asked ChatGPT to “make a productivity app.” And if that obsessed person has built an audience on social media with said specific knowledge, you have successfully broken past the 5th level of market sophistication into the land of irreplaceability (for now).

Your years of reading, experimenting, failing, and figuring things out is the moat.

The irony is that AI is making human knowledge more valuable, not less.

AI made everyone fast. Anyone can make a course. Anyone can write content. Anyone can copy anyone. But very few people are uncopyable. Very few people have spent years developing so much fine-tuned experience (taste) that nobody would think to copy the details that make it work.

Build the thing only you would think to build.

– Dan

If you want to watch the video or podcast version of this letter, you can do so on YouTube or Spotify here.

Here are related letters if you want to continue reading: