Making money with AI: the honest math and my experience

The uncomfortable truth first

AI does not earn money. Your business earns money — AI only changes how much of it stays with you and how fast it comes about. That is the most honest answer to the question of how you can make money with AI.

So no wrong impression arises: Very good money can be made with AI. Just not the way you are being told. In this article I take away the promise, not the opportunity.

I say it this clearly because right now a whole industry is selling the opposite: start this AI side hustle, build that faceless channel, let “the AI work for you” while you sleep. Whoever follows such promises starts from zero and competes with thousands of others for the same copied idea.

Meanwhile you are sitting on something none of these sellers has: a running business with real clients, real processes and real expertise. That is where the first and biggest money sits — that is page one of this article, with three calculations to check for yourself.

And the window for it is wide open right now. According to the Bitkom survey for 2026, 41 percent of German companies actively use AI — a year earlier it was 17 percent. Among small and medium-sized businesses, though, it is only 25 percent so far. Only companies with 20 or more employees were surveyed; the number says nothing about solo self-employed people, which makes the head start for you bigger rather than smaller.

But there is a second side, and I am not leaving it out just because it is inconvenient for my own thesis: the market has changed. Having only a single income leg is simply too dangerous today. AI makes second income streams possible that previously failed on time, technology or budget. There is a serious approach to that too — it is in the second half.

The overview: your own business first, the side projects after

Most people want to start at 3 or 4 because it sounds more glamorous. The order with the better numbers is 1, 2, 3 — and 4 only once you know what you can carry.

Now one by one — with numbers.

Calculation 1: what you no longer spend

The fastest way to make money with AI is unspectacular: replacing expenses.

My new website — including the blog you are reading this on — came about entirely in dialogue with Claude. No agency, no site-builder subscription. As a non-techie, I built my first app in a good week — a few years ago that would have taken a developer team and a budget. And the articles on this blog come out of a system that does the work of a small editorial team.

Run the numbers for your business, in three lines:

1. Contractors: What have you spent in the last twelve months on texts, graphics, landing pages, small tech jobs? Even a single outsourced job usually costs more than a year of AI tooling.
2. Tool subscriptions: How many subscriptions do you pay that overlap in function — or that you keep out of habit? Well-led AI often replaces several of them at once.
3. Waiting costs: The most hidden item. Every week of waiting for a contractor is a week in which an offer is not selling, a page is not live.

Money saved is money earned — just without the VAT return.

Calculation 2: your owner’s pay

The second calculation is missed by almost everyone, because it does not appear on the bank statement: your hours.

Let us make it concrete — with your numbers, not mine. Count the hours per week that flow into work a well-led AI system can take over: draft texts, summaries, evaluations, research, standard emails. Then move the two sliders:

For most solopreneurs I work with, five hours a week is a conservative estimate. And if you just hesitated at the hourly rate: do not enter what you dare to charge, but what your hour really brings in when you give it to client work. I made that mistake myself for years.

These hours do not disappear — they become free. For client work that brings revenue. For the offer that has been half-finished for months. Or, and that too is a business decision: for the evening off. “Making money with AI” here means: stop giving away your owner’s pay to routine work.

And this calculation has a second punchline hardly anyone says out loud: Today you can offer services you would previously have needed a team for. The consultant who suddenly delivers the implementation along with the advice. The copywriter who no longer hands over just the text but the finished landing page. Whoever leads with AI instead of just typing sells bigger packages to the same clients — without hiring a single person.

Calculation 3: the product already sitting inside your business

The third calculation is the most exciting. When AI radically cheapens implementation, products become possible that previously failed on cost or technology — out of the expertise you have long had.

My example: a question my community had been asking me for months became a finished product in a good week — app, payment system, legal texts, landing page. The first payment came via a single story with around 40 views. Not because I performed marketing magic, but because the problem was real: my clients had been shouting it at me, I listened and delivered.

The calculation behind it: a product like this used to require an investment of team, budget and months — a risk most solo businesses rightly do not take. Today the stake is one week of guided work, almost no starting capital. If the product carries, you have a new income source. If not, you have spent a week learning what your market does not need — at the price of zero outsourced jobs.

What problem have your clients been shouting at you for months? That is your product idea — not the twentieth “AI money machine” course from someone who earns their money with the course and not with the machine.

My experience with AI: the balance from six weeks

So this does not stay theory, I am laying my own experience open. Over six and a half weeks I measured what came into being in my business with AI — every session, every working hour, every euro spent.

In that time, 19 projects were finished. Among them the product from calculation 3, which has been selling since day one. Three reworked websites including findability on search engines and AI answer machines. A blog brought up to standard. Commissioned conventionally — at market day rates — that would have been around 346 person-days.

And now the part that matters more to me than the big number: My first calculation was wrong. €505 in tool costs against a conventional equivalent of around €400,000 — factor 800. That sounds terrific, and exactly such numbers are what you read everywhere. It is still dishonest, for two reasons.

First: My time is not free. I myself spent around 89 hours in those weeks leading, checking and deciding. And here I deliberately do not use what I would pay a contractor, but what that hour really costs me: it was lost to client work. Calculated at €450 per hour, those 89 hours come to a good €40,000. They belong on the cost side — otherwise I am sugar-coating my own owner’s pay, the very thing I just warned you about in calculation 2.

Second: I would never have commissioned all of it. Four of the 19 projects I only did because they suddenly became possible. I must not count those as savings — I only saved what I would otherwise genuinely have bought.

The honest balance looks like this: around €280,000 in jobs I would actually have commissioned, against a good €40,000 in full costs. Factor 7. Not 800. And this factor 7 is the real news — because it withstands every follow-up question. I charged myself the most expensive hour I have and struck out everything I would never have bought anyway. What remains is still: for every euro this cost me, seven euros of job value came into being.

Why I am telling you about the correction: that is exactly how you recognise honest AI math. Whoever shows you factor 800 is not counting their own time and pretending they would otherwise have bought every project at full price. Do your own math along the honest pattern — full costs including your hours at their real value, savings only for expenses that would have become real. A single-digit factor that holds is worth more than a three-digit one you cannot explain to anyone.

The second side: making money with AI alongside your business

Up to here, this was the calculation inside your own business. Now the other side — the one that always comes first in the hustle videos and that I am still not brushing away: parallel income streams.

Because one thing really has changed: what used to require its own team, a second budget and a second working day is now doable for one person. Affiliate sites that AI builds and maintains while you only steer them. Channels on TikTok, Instagram or YouTube whose production a guided system takes over. Small digital products running beside the core business. I am not talking from hearsay — I am currently building a portfolio of small apps and affiliate sites myself and testing video channels, parallel to my consulting business.

And this is not a niche for a few twenty-year-olds. According to a survey by Lexware and WeCreate, around 3.5 million people in Germany work on content-based business models. Goldman Sachs expects the creator economy to grow worldwide to around 480 billion dollars by 2027 — from about 250 billion. There is money there, and plenty of it.

It is just earned differently than the advertising claims. And the uncomfortable numbers from the same German survey tell you fairly precisely how: 62 percent of the creators surveyed report strongly fluctuating income, and reaching financial independence takes on average around two and a half years. Those surveyed were creators with large reach — so for a start you should plan for more time rather than less.

That is still not a counter-argument. That is the building manual. Whoever expects fluctuating income builds several streams instead of one from the outset — and lands exactly where this section is heading. Whoever plans for two and a half years does not quit disappointed after three months. What people fail at is rarely the market. It is the timetable someone sold them.

Before you start any of this, answer a single question, in writing: What am I doing this for — for my brand or for money?

It sounds banal but decides everything. A channel meant to make your brand known is run differently from one meant to bring income: different content, different frequency, different success metrics. Whoever wants both at once from the same project usually gets neither. My recommendation after 26 years of self-employment: Do both — but separately. The focus stays on your business; that is where calculations 1 to 3 sit. And in parallel you soberly check which streams to the left and right make sense for you.

“Soberly check” means: three carrying-capacity questions, answered honestly —

1. What can I carry technically? Not what is theoretically possible — but what you can keep leading with your current knowledge without it paralysing you.
2. What can I carry financially? Every stream costs before it delivers: tools, domains, ramp-up time. Calculate the stake before you dream of the return.
3. What can I carry time-wise? The hardest of the three. A neglected second leg is not an income stream but an open construction site draining energy.

If one of the three answers is “not really”, that is not failure — it is prioritisation. Then your energy belongs one hundred percent in calculations 1 to 3, and the second side comes back on the table later.

Why I take the topic this seriously at all: The market has changed. Jobs that came as a matter of course for twenty years no longer come as a matter of course. Platforms change their rules overnight. Having only a single leg today is simply too dangerous — not as a panic message, but as a risk calculation. Two or three deliberately led streams are the insurance that, with AI, you can afford for the first time without tearing yourself apart.

The reality check: what does NOT work

So this article keeps what “honest math” promises, here is the counter-list too. Be sceptical of anything that sounds like this:

• “Passive income with AI — money while you sleep.” Usually only the seller of the promise earns. Systems can scale, yes — affiliate sites too, channels too. But they do not come into being passively, and they do not run without leadership.
• “Mass content, a hundred videos a day, the AI does everything.” Volume without a concept competes with thousands of identical offerings — and platforms increasingly downgrade exactly this content. A guided channel with a clear goal is the opposite of that.
• “You only need this one tool / these 10 prompts.” Prompts without changed thinking achieve nothing. If it were that simple, everyone would do it — and then it would no longer be worth anything.

The common denominator of these promises is not where the money is supposed to be earned — but how: without leadership, without your own math, without a real problem being solved. That is exactly what separates the guided parallel stream from the last section from the hustle: you have answered the what-for question, done the carrying-capacity math and treat the second leg like a business — not like a lottery ticket.

And the honest footnote missing from the hustle sellers: none of these calculations works without structure. Whoever does not know their business — which processes are standard, what only you can decide — only automates their chaos. Structure before tools. That is also why I believe the future belongs to solopreneurs: one person with leadership and tooling today moves what used to require an apparatus.

How to find your biggest calculation

Four questions, in this order:

1. What have I paid contractors and tools in the last 12 months? (→ Calculation 1: that is where your immediate money sits.)
2. Which 5 hours of my week could a good employee take over? (→ Calculation 2: that is your owner’s pay you are giving away.)
3. What problem have my clients been shouting at me for months? (→ Calculation 3: that is your dormant product.)
4. What do I want a second stream for — brand or money — and what can I carry technically, financially and time-wise? (→ The second side: your income-leg check.)

If you would rather not go through this alone: that is exactly what I built the location check for — a few minutes, honest questions, and you know where in your business the AI money really sits and what your most sensible first step is.

Frequently asked questions

Can you really make money with AI?

Yes — but differently than the advertising promises. AI does not earn money on its own; it makes your existing business more profitable: through replaced contractor costs, freed-up working hours and products that were previously too expensive to build. And it makes second income streams leadable that used to require a team.

Is making money with AI legitimate?

It becomes legitimate where the math withstands every follow-up question: your own working time sits on the cost side, and only what you would genuinely have spent otherwise counts as savings. It becomes illegitimate with promises of no effort — “money while you sleep”, three-digit factors, courses about money machines that earn their money with the course. One test sentence that almost always suffices: whoever does not count their own hours is not counting.

How do I make money with AI fastest as a self-employed person?

Via the expenses calculation: replace the next contractor job (text, landing page, evaluation) with well-led AI work. That works immediately and without risk — and even a single replaced job pays for the tooling for months.

How much money can I save or earn with AI?

Calculate with your own numbers: hourly rate × routine hours per week + contractor expenses of the last twelve months. The first part is worked out for you by the calculator further up in the article — two sliders, an annual value immediately. My own honestly calculated balance from six weeks is also above: factor 7 between full costs and what I would otherwise genuinely have spent.

Should I build a second income leg with AI alongside my business?

First answer the what-for question: is it about your brand or about money? Then the three carrying-capacity questions — technical, financial, time. If all three hold: yes, deliberately and guided. The market has changed; only one leg today is a real risk. But the focus stays on your core business — that is where your head start is uncatchable.

What about passive income through AI?

Be sceptical when someone sells you “money while you sleep” with AI — usually only the seller of the promise earns there. Affiliate sites and channels can very well bring income, but they do not come about passively: they need building, leadership and regular decisions. Scalable yes — passive no.

What is the best AI for making money?

The question matters less than the advertising claims: one tool, properly led, beats ten tools side by side. I work with Claude — what it costs is in the article “What does Claude cost?” on this blog. More important than the tool choice is taking stock of your processes.

What experiences are there with making money with AI?

My own are laid open in this article: six weeks measured, 19 finished projects, €505 in tool costs and 89 hours of my own work. Honestly calculated, that yields factor 7 — not the factor 800 my first, wrong calculation produced. That is exactly why experience reports on the internet diverge so widely: the big numbers almost always come from leaving out one’s own working time.

Can you make money with AI without starting capital?

In your own business, yes. The first calculation in this article is an expenses calculation — it needs no capital, just a job you no longer outsource. For a second leg beside it you also need hardly any money, but time, and for the self-employed that is usually the scarcer currency. Whoever reads “without starting capital” and understands “without effort” will be disappointed.