How to Make Money With AI Without Trading Your Time (The Toll Booth Model)

If you want to make money with AI without trading your time, you will hit the same wall every founder hits. The AI saves you hours, and then you fill those hours with more client work, and your income is still a function of how many hours you are awake. The way out has less to do with automating your calendar and more to do with changing where the money comes from. Build something that collects on volume, and your income stops being a headcount problem.

I got there in a conversation this week with a friend who builds AI harnesses for a living. They run a creative services business that has paid them well for ten years, and they described it as pure hunting. An animal comes out of the woods and they shoot it. That worked beautifully and then it stopped, more than once, for months at a stretch, because there was never a system underneath it that could call an animal out.

Then they said the line the whole conversation turned on. Making money for your time is a dead-end road.

Why Is Trading Time for Money a Dead End?

Because presence is the constraint, and presence has a hard ceiling.

They laid it out plainly. The problem with coaching and with client production work is that they have to be present for it. If they get sick, they are done. When they get old, what then. Did they collect enough money on the way through. That is the whole risk profile of a business built on being in the room.

I had almost the same conversation the morning before with a friend who coaches. I asked them a hypothetical. If you were not allowed to make money as a coach, if trading your time was off the table entirely, what would you do. They thought about it for a while and said coaching. I told them that was the one answer they could not give, and they went straight back to it.

That is the trap most of us are in. The work we love is the work that requires us to show up, which means every dollar has a person-shaped bottleneck attached to it. AI does not fix that on its own. Handing an assistant to someone who bills by the hour just gives them a faster way to bill by the hour. If that pattern sounds familiar, running a one-person business with AI is the version of this problem most solo operators solve first.

How Do You Make Money With AI Without Trading Your Time?

You make money with AI without trading your time by owning a position that collects a fraction of something that happens a lot.

They called it a toll booth. Their example was a routing service that sits between developers and the AI models they call, collecting on traffic it does not own. Then they gave the older version of the same idea. Who won the gold rush. The people selling shovels.

Their compression of it was two words. Data and processing power. Those are the two things everything else in this cycle has to buy.

I heard the phrase and immediately used it as a filter. We do not ship anything unless it runs like a toll booth. That single question kills a surprising number of ideas before they eat a month.

Here is my own first attempt at it. Cloudflare added a monetization gateway that lets you attach a wallet to your site and charge AI crawlers fractions of a penny to read your content. I built a penny machine against it as soon as I understood it. Am I aiming at hundreds of thousands. Yes. Would I take pennies in the meantime. Also yes. The point is that the position exists now, the content that feeds it is content I was producing anyway, and the meter runs whether I am at my desk or asleep.

What Are the Best AI Business Models for Solopreneurs?

Three shapes are working right now for people operating alone.

The toll. You sit between two parties who are going to transact regardless and you take a sliver. Data access, routing, aggregation, marketplace. Your revenue tracks volume rather than hours.

The published asset. You produce something once and it works while you sleep. My blog is this. A Claude Code routine takes a conversation like the one I am describing, strips the names, checks it against the 250 AEO keyphrases I keep in Notion, drafts the post, and publishes it every night. In the last thirty days that produced 5,200 organic views and 3,800 visitors to my site with zero additional hours from me.

Now hold that against honest math. Put a popup on it and maybe a tenth of a percent sign up. A tenth of those go further. Maybe the whole thing produces one client in a year. That is a real number and it is still worth it, because the system runs at zero marginal cost forever and every conversation I was already having makes it bigger.

The narrow product. More on this below, and it is the one most people get wrong.

If you want the broader architecture around all three, building an AI-first business as a solopreneur covers the operating layer that sits underneath any of them.

Should You Build One Big AI Product or Many Small Ones?

Many small ones, and make each one embarrassingly specific.

Their argument was that you cannot sell people a horizontal thing that could do anything. The pitch has no edge and the buyer has nothing to picture. You build a vertical instead. This one picks up your keys, even though the underlying engine could drive the car. Little harnesses that run a few skills over and over, aimed at one job somebody already knows they hate.

Then you market the daylights out of them and accept that most will go nowhere.

They put a number on the acceptance part. Babe Ruth hit a home run about eight and a half percent of the time. That is the swing rate you are signing up for, and it only works if the cost of one swing is low. Which is the actual argument for building small.

My own version of low-cost swings is architectural. I deliberately avoid building one giant chain, because a long chain breaks in one place and takes everything with it. I break the work into small pieces that each stand alone. One routine turns a transcript into a post. Another checks whether it is ready. Another publishes it. Any one of them can fail on a Tuesday without costing me the other fourteen. Agentic AI for a small business is worth understanding before you decide how many pieces you want.

Can You Make Money With AI Without Trading Your Time as a Coach or Consultant?

Yes, and the honest path runs through your own delivery before it runs through a product.

Start by turning what you already say into an asset that keeps working. Every call you have is source material for something durable, and most consultants throw all of it away. Then move the delivery itself from one person to many, which is the same lever a coach pulls when they scale a coaching business with AI.

The toll comes later, and it usually shows up as something you built for yourself first. My publishing system was built because I wanted the traffic. The penny machine sits on top of it because the traffic turned out to be worth something to somebody else. That order matters. Build for your own use, prove it works on you, then look at who else pays to pass through it.

There is a trap here worth naming, because I walked into it. I got an always-on agent setup running on a machine next to my desk about four months ago. I was burning around ten dollars a day in tokens, the cron jobs kept breaking, and I was losing forty-seven minutes at a time banging my head against the wall over a thing that was producing exactly zero new business. Efficiency work feels like progress. Get out of anything that costs money and time while making neither. The related question of using AI to replace a $50,000 hire is a better use of the same energy, because at least the output has a price attached to it.

How Long Does It Take Before an AI System Makes Money?

Longer than a launch and shorter than a career.

They told me about a stretch of focused work on search visibility for a specific type of client. It took six months to produce anything. Then it produced. Six months of nothing is the part nobody puts in the case study.

Being early is the whole edge, and the analogy they used has stuck with me. When digital cameras arrived, a photographer they knew kept charging the film-era rate per event while shooting as many frames as they wanted at no marginal cost. That photographer made a killing for a few years. Then everybody knew, and the pricing corrected. We are inside that window right now with AI, and the window closes when the tooling gets good enough that none of this requires taste.

So the timeline that matters is whether you are building during the years when building is still an advantage. When your thing pays matters far less than that.

What to Do This Week

1. Write down every way money currently reaches you and mark which ones require you to be present. That is your exposure.
2. Pick one thing you already produce and put it somewhere it can keep working after you stop touching it.
3. Ask the toll booth question about your next idea. Does it collect on volume, or does it bill for hours.
4. Choose one embarrassingly narrow product instead of one flexible platform. Give it one job.
5. Cap the cost of a swing. If an experiment cannot fail cheaply, redesign it before you start.
6. Audit anything running right now that costs money and produces no revenue. Kill it or give it a deadline.

The Part Worth Sitting With

Two things came out of that conversation that I keep returning to.

The first is that being early is a real asset and it expires. Both of us look around and see the same thing. Most people are still typing into a chat window, and building agentically is rare. That gap is worth money for a limited number of years.

The second is that zero to one is the whole game right now. My architecture is not elegant. This entire category barely existed a moment ago. Go zero to one, then do it again. You get to place this bet in week one, and that is rarer than being right.