How to Make Money with AI
by Isaac Otu
Chapter 1
A small business owner needs a chatbot. Not a big one, not a research project. Just something that can answer "what are your hours" and "do you deliver" on the website at midnight, without a human on the other end.
Ten years ago, that request went to a software team. There was a scoping call, a few weeks of development, and a bill with several zeros on it. Today it goes to a solo operator running a no-code AI platform. That person builds the bot in an afternoon. The price runs somewhere in the neighborhood of $300 to $1,000 to set it up, plus $100 to $500 a month to keep it fed with current information and fix whatever breaks. That is not a made-up number. It is the actual, published pricing pattern one AI chatbot platform quotes to the people building on top of it, for exactly this job.
Nothing about that chatbot is a new kind of business. Someone builds a small piece of software, sells it to a business that needs it, and charges for setup and upkeep. People have done that since software existed. What changed is who can do it, how fast, and for how little. The team shrank from several developers to one person. The timeline shrank from weeks to an afternoon. The price shrank from a number with several zeros to a few hundred dollars. Nothing here was invented. Something was compressed.
That is the whole claim this book makes, in five different markets, from five different angles, until you can spot it yourself without help. AI has not invented new ways to make money. It has compressed the time, the team size, and the skill that a handful of already-existing ways to make money used to require. The money still goes to whoever supplies the part the software can't. Knowing which chatbot a specific business actually needs. Spotting which three hours of someone's week are worth automating. Deciding what a client should hear, and how to say it. The compression is real. The judgment that decides what to compress is still entirely yours to supply. It is still what gets paid.
This book covers five categories of AI-enabled income. Every one of them existed before a large language model ever answered a question.
Freelancing existed the moment someone with a skill first sold it by the job instead of the year. What AI compressed is the time between having the skill and delivering the job. A piece of freelance writing, design, or development work that used to take days now often takes hours. A first draft, a first mockup, or a first working version arrives fast enough that the freelancer's real job becomes judgment and revision, not production from a blank page.
Small AI-powered products existed the moment someone first built a tool to solve a specific problem and charged people to use it. What AI compressed is the size of the team required to do that. A working app, a browser extension, or an automated workflow used to need a founder, at least one developer, and usually months. It can now be built by one person, in a fraction of that time, provided that person has correctly identified a real, specific, already-felt problem someone will actually pay to have solved.
Automation consulting existed the moment a business first paid an outside expert to make some part of its operation less manual. What AI compressed is the cost and speed of building the automation once the diagnosis is made. Connecting a business's existing tools together, so a lead that comes in by email ends up logged, answered, and followed up on without anyone re-typing it three times, used to be a custom engineering project. One competent person can now wire it together with off-the-shelf platforms like Zapier, Make.com, or n8n, in days, without writing custom code for most of it. The real, published pricing for this work runs from a few thousand dollars for a defined one-time build to a few thousand dollars a month for ongoing management. Some agencies add an annual maintenance fee on top, worth a fifth to a third of the original project cost. The billable skill moved from writing the automation to correctly diagnosing which of a client's tasks are actually worth automating, and being trusted enough that the client believes the answer.