How to Build Recurring Revenue With AI: The Penny Machine Approach

How to build recurring revenue with AI is the question sitting under every conversation I have with people who use AI at the highest level. You can automate your content, run agents around the clock, build a product in a weekend, and still wake up feeling like the recurring part never arrived. This post is the honest version of that problem and the approach that finally got my gears turning: the penny machine.

I spent an hour this week on a call with a friend who is one of only three people I know who operate with AI the way I do. He told me he wakes up at 3 AM every day with a gut-wrenching feeling that he is behind the curve. Then around 7 AM he starts meeting clients and peers and realizes the rest of the world is way behind him. Every day, seven days a week. I wake up at 5 AM with the same feeling. We both know it is about one thing. Four years all in, and the recurring revenue has not caught up to the skill.

How to Build Recurring Revenue With AI When You Already Use It Every Day?

Because the thing you are best at, building, is the thing that got commoditized first.

My friend is a product builder. He has played every role in shipping software. His pattern for the last two years: a brilliant idea, holy shit nobody is doing this, he builds it, gets traction, gets ready to deploy, and twenty people already have it out there. The tools are in everyone's pocket. The knowledge of how to build is in everyone's pocket. Marketing and positioning are one prompt away for anyone.

His conclusion was to stop building unless the opportunity is extremely hard to do. Capital. Network access. Domain knowledge. Five pieces you cannot assemble without expertise. High barrier opportunities deserve a build cycle. Everything else is a race you enter with twenty people who are already ahead.

I asked him a question on that call that reframed the whole hour. Would we be feeling any of this if the way we create with AI was generating 6,000 dollars a day? That is about two million a year. His answer was honest. Some of the emotion would subside, and a new one would show up: protecting the thing, because someone could replicate it and 6,000 becomes 600.

So the recurring revenue question has two halves. What can you build that cannot be copied, and how do you get paid in a way that does not depend on the customer winning?

What Is the Penny Machine?

The penny machine is the smallest possible version of recurring revenue with AI, and it exists to change the question you are asking.

Everyone in my world is trying to think of the magic 6,000-dollar-a-day idea. I broke revenue down by how hard each unit is to earn. Will someone pay me 100,000? Ten thousand? Five hundred? Forty-nine dollars? A penny? Every step down is easier to earn and easier to automate.

The penny machine asks one question. How do we get AI to generate our first penny without us?

The moment I started thinking this way was at a resort pool over a holiday. Hundreds of people, shirts off, drinks in hand, and I was probably the only person there thinking: they are monetizing every one of these people right now. We tell ourselves there is no money in the market. These people were spending money in front of me. I deserve to be able to earn from them too, and the lowest unit I could earn is a fraction of a penny.

Start there. One penny earned by AI with zero human involvement is proof the machine works. Then you scale the machine instead of scaling yourself. If you want the honest math on whether an AI build is paying for itself, my post on how to measure ROI of AI in my business is the companion to this one.

What Is a Toll Booth Business Model in the AI Era?

A toll booth gets paid whether the market is good or bad.

Someone I shared the penny machine with said, oh, this is the toll booth concept, and then asked me who made the most money off the legalization of marijuana. I had no idea. The companies that sold the lights. Whether your grow house succeeds or fails, you needed lights to start. They got paid either way.

So much of consulting and coaching lives on the other side of that. Do you have the budget? Do you have the time? Can you get out of your own way this quarter? A toll booth removes those questions. Boop, you are through. Boop, you are through.

The specific toll booth that lit me up came from my hosting provider. Cloudflare released a monetization gateway that lets you charge AI agents for access to your content and data. Read that again. There are now more AI agents on the internet than humans, and they are scaling faster than any audience you will ever build. Marketing to agents, what I call answer engine optimization, is a fraction of a fraction of a fraction of the market right now, which means it is uncontested.

What agents find valuable and what they will pay for is still an open question for me. What I do know is that if you and I are among the most advanced AI users on the planet, we can reverse engineer what we would want and build the toll booth for it. My starting point was how I optimized my entire website for AI search, and the gateway is the next step past it.

What Assets Cannot Be Commoditized When Everyone Has the Same AI Tools?

Three things. Your conversations, your archive, and your judgment.

Your conversations. On the call I said something out loud that stopped us both. What if you and I are generating recurring revenue every single time we talk? Two people at this level, with a script of questions by design, seeding an AI with context nobody else in the world can provide. That is not commoditized. Only the two of us can do it. Everything we research has to be assumed copied by someone else in the top fraction of a percent. Our conversations cannot be.

Your archive. I have written down every conversation I have had for years. Every course, every book, every podcast, every client call. AI does not need an encyclopedia. It already has one. What it needs is the experience, knowledge, and values of the person it is operating with, and that is the thing most people cannot hand it because they never wrote it down. This is the whole reason I tell every founder to build a second brain with AI before they build anything else.

Your judgment. My friend put it better than I could. A kid out of college has the same access to the same tools and may be more fluent in them. What they are missing is the domain knowledge. Twenty years of trials, failures, and successes is what lets you look at the best answer AI can give and say, this part is right, tweak this, remove that. Without an expert next to it, the output is inference all the way down.

One more that surprised me. Anchor the why. Asking an AI to generate 6,000 dollars a day is one prompt. Asking it to generate 6,000 dollars a day so I can take my son out of school and design an entrepreneurship curriculum for a five-year-old is a different prompt. The purpose becomes the anchor point every idea has to align to. My friend's read: that context is the grounding, and then the AI has the freedom to find what we are not seeing.

How to Build Recurring Revenue With AI Starting From One Penny?

This is how to build recurring revenue with AI when you refuse to compete on ideas.

1. Break revenue into units. List what someone would pay you: 100,000, 10,000, 500, 49, one penny. Design for the penny first.
2. Find your toll booth. Where can you get paid regardless of the customer's outcome? What do people need to even start, the way every grow house needed lights?
3. Audit what cannot be copied. Your conversations, your archive, your judgment, your network. Everything else, assume twenty people already shipped it.
4. Anchor the why. Give the machine the purpose behind the number so every idea it generates has something to align to.
5. Point the machine at two metrics. Audience and revenue. My friend and I both have machines that can run 24 hours a day. If they are not growing a channel, a list, an answer engine footprint, or revenue, we are failing them or they are failing us. Distribution is the whole game because anyone can build.
6. Use demand data, not instinct. Search Console, video research tools, trend data. Find where demand is high and supply is low, including the boring industries nobody in AI is looking at.

The path from a coaching business to a penny machine runs through agents doing the work. If that is new territory, start with what agentic AI is and how a small business owner can actually use it.

How Much Should You Trust AI Agents to Run Revenue on Their Own?

Less than the marketing says, and more every month you watch them.

My friend runs a business manager bot for each of his clients. It has every file, every meeting note, every transcript. Every night at five it pings him: any updates? Nothing today, pass. Oh, we had a meeting, it is in the transcript. Two weeks later the status report is written and waiting for him to push. Hours of tedious work gone, per client.

He chose that design on purpose. He wants to skip the chief of staff bot that tells other bots what to do while he watches from a distance. He wants visibility. He wants to interact with it directly. His rule: chunk the tasks by design so you can validate and verify. If it runs spot-on for six months, then he will think about letting another bot manage it.

We both built armies of agents earlier this year. Fifty of them, swim lanes, teams, orchestration. We both abandoned them for the same reason. We were spending more time fixing what broke than creating value, and no revenue came out the other end. The lesson was simple. Complexity is the enemy of greatness. Simple is hard because simple scales.

That is the trust ladder for revenue with AI. Chunk it. Watch it. Verify it. Earn autonomy the way you would with a new hire. My own version of this path is in how I built AI agents as a non-technical founder, and the routine-first approach there is the same idea.

We ended the call with one reframe. My friend said his whole job now is taking complexity and simplifying it. I said, change one word. Taking revenue and simplifying it. That is how to build recurring revenue with AI without competing on ideas. Build the penny machine. Watch it earn the first penny without you. Then scale the machine, and go sit somewhere fun while it runs.