I am going to tell you something that still sounds a little unreal when I say it out loud. My team is on pace to launch one brand new software app every single month this year. And we are doing it with two people and an AI.
That was the goal I set, and it is also insane to type. But it is happening. Since January we have shipped several new apps, plus a full desktop and web version of our first one, and there are more on the way. Roughly one a month.
The point of this article is not to brag about the count. It is to show you the operating model underneath it, because the model is the interesting part. A two-person team plus AI is a genuinely new way to build a company, and I think a lot more businesses are about to run this way. So here is exactly how it works.
The two-person plus AI model
Let me be specific about who does what, because the magic is in the split.
We have one full-time developer, Meri. She handles the heavy architecture, the hard decisions, the parts of a software product that genuinely require a skilled engineer. That work still needs a real person, and she is excellent at it.
Then there is AI, which fills in everything around her. It writes features. It handles the routine building. It ships faster than I ever thought possible. It is genuinely working like a part-time developer sitting next to her.
And there is me, setting the direction, deciding what we build, and increasingly building the simpler pieces myself with AI tools because I no longer need to hand every small thing to a developer.
Here is the key idea, and I want to be clear about it. The AI does not replace Meri. It multiplies her. It takes the routine work off her plate so she can spend her time on the parts only a senior engineer should touch. One skilled human plus AI does the work that used to take a whole team. That is the whole model in one sentence.
If you want the bigger argument for why this favors owners and small teams instead of threatening them, I laid it out in AI is replacing employees, not owners.
My actual AI stack
People always want to know the specific tools, so here is how I actually use AI day to day. Think of it as three layers.
The first layer is a daily assistant. I use it as a co-pilot for everything from research to drafting content. This newsletter you are reading? Drafted with it. It is the thing I reach for first thing in the morning.
The second layer is automation. I have AI running recurring jobs across my whole business: SEO audits, content pipelines, and weekly reports across my entire portfolio of 14 sites. These are the boring, repetitive jobs that used to either eat hours or simply not get done. Now they run on their own and hand me the results.
The third layer is coding. This is the part that works like a part-time developer, writing features and shipping alongside Meri. It is the reason a two-person team can keep a one-app-a-month pace.
Each layer on its own is useful. Stacked together, they let a tiny team operate like a much bigger one. And it reaches past software. In one case I used AI to research state law and draft the arguments that helped me push back on a local zoning interpretation that was blocking one of my building projects. A city eventually adjusted its code. The tools that write my features also research my legal questions and run my marketing reports. That breadth is the real advantage.
The result that keeps me up at night
Now the part that genuinely reframes everything, because a model is only as good as what it produces.
One of our apps helps real estate investors track the hours they need for a specific tax status. We built it for under 300 dollars. It now produces five figures every single month, with no marketing spend.
Sit with that number. Under 300 dollars to build. Five figures a month, every month, since.
But here is the line that really got me. That app is now out-earning some of the actual rental properties I own. Let me say that again. A 300 dollar app, with no tenants, no maintenance calls, and no property taxes, is generating more monthly income than doors I bought and manage.
I have spent years building rental income. It is real and I love it. But a rental has tenants, repairs, insurance, vacancy, and taxes. This app has none of that. It is truly passive recurring revenue that shows up while I sleep. No toilets. No tenants. Just margin.
That is not a fluke of one app. It is what happens when you point the same asset-building framework I use on rentals at software instead. Build the asset once, put systems around it, get paid over and over. I made the full case for why a real estate investor is unusually well suited to this in why a real estate investor should build apps.
Why the cadence matters
A lot of people fixate on the fact that any single app might not work. That is true, and it is also the point.
When it cost 15 thousand dollars and two months to build a product, every launch was a big bet. You had to be right, because being wrong was expensive. So you launched rarely and agonized over each one.
When a two-person team plus AI can ship a new app roughly every month, the math flips. Each launch is cheap. You are not betting the business on any one of them. You are taking a lot of small swings and letting the market tell you which ones connect. One app that produces five figures a month easily pays for a dozen that go nowhere.
That is the quiet advantage of the cadence. It is not that every app is a winner. It is that shipping often and cheap means you only need a few winners, and you find them by shipping, not by guessing. I use AI build tools for the fast, simple version of this and pair them with a real developer for the hard stuff, which I broke down in how I build real software without writing code.
What this means for your business
You might not want to build software. That is fine. The lesson here is bigger than apps.
The old rule was that doing more required hiring more. More output meant more headcount, more payroll, more overhead. That rule is breaking. A small, skilled team with a good AI stack can now produce what used to require a department.
So the question worth asking about your own business is not "what can I do myself." It is "what could my small team do if AI handled the routine layer of every job." The reports that do not get run. The content that does not get written. The features that sit in a backlog. The research you keep meaning to do. Much of that can now run on an AI layer while your people focus on the work that actually needs a human.
The businesses that figure this out are not going to hire their way to scale. They are going to build a lean team, hand the repetitive work to AI, and quietly out-produce competitors ten times their size. If you are building that lean team, my approach to the human side is in how to buy back your time by hiring a remote team.
If you want to start your own version of this, do not try to build all three layers at once. Pick the single most annoying repetitive job in your week, the report you dread or the content that never gets written, and hand just that one job to an AI tool. Get it working, trust it, then add the next one. Within a few months you will have quietly built a stack that runs the boring half of your business on its own, and you will wonder how you ever did it by hand.
The takeaway
Two people and an AI, shipping a new software app roughly every month, with one of those apps out-earning rental properties I own. A few years ago every part of that sentence would have sounded like a stretch. Now it is just how we work.
The model is not complicated. Keep the team small and skilled. Give the hard, human work to a real expert. Hand the routine work to AI across coding, operations, research, and writing. Then ship often, keep the cost of each try low, and let the winners pay for everything else.
You do not need a big team to build big things anymore. You need a small one that knows how to use the tools. That is the whole point, and it is the most exciting time I have seen to build.
This article shares my own experience building software with a small team and AI, and is not financial, legal, or investment advice. Results vary widely, and most software products do not become profitable, so treat any build as an experiment, not a guaranteed return.
