This month we are using AI to build an entire business. From idea to sales, every step will be AI and you'll get to see every prompt or agent I use along the way.
It starts today!
I used Claude to scour the Internet for business ideas and boy did Claude bring home the bacon! Claude found me a niche that pays $15k - $35k per sale 🤯
Have I told you yet that Claude is my homeboy?!?! Seriously, my $200/mo Max plan is the most profitable spend out of all my business expenses. I'd pay $2,000/mo and it would still be profitable.
Ok, so let's get into it.
How to Find Lucrative Niches
We started with one rule: whatever business we build this month has to be a newsletter business.
Why a newsletter? Because I need AI to be able to do ALL the work. Not some of it. All of it.
Opus, Fable, and GPT 5+ level models are scary good at writing newsletter content and researching markets. And thanks to the Kit MCP (an MCP is a connector that lets Claude log into your software and click the buttons for you), Claude can even set up the email drafts inside my Kit account (the same email platform this newsletter runs on).
All I have to do is press send. AI can't do that part for me. Yet.
So last Thursday afternoon I opened Claude and gave it this exact prompt:
Embody an elite Internet research assistant. You have decades of experience meticulously scouring the web to create comprehensive research reports for your clients. You use Google dorks, directories, social media, /last30days skill, and a vast array of clever web apps to help you find even the tiniest unknown corners of the web.
For this task I want you to create a comprehensive report on the most successful email newsletters of all time, most popular niches for newsletters, most profitable niches for newsletters, worst niches for newsletters, and great newsletter niches with low competition and high earning potential.
GOAL: Find the best niches for starting an email newsletter in 2026 and beyond.
Claude came back with a full report. The most successful newsletters ever built. The niches everyone piles into. The niches that quietly print money.
And the graveyard.
The graveyard was my favorite part. Claude ruled out daily news (The Messenger torched $50 million in 8 months and died). It ruled out stock picks and gambling picks (the FTC has handed out multi-million dollar fines in those lanes). It ruled out crypto (your newsletter lives and dies with token prices), celebrity gossip (advertisers blocklist the whole category), and generic motivation content (NewsGuard counts over 3,000 AI content farms already pumping that slop out).
The pattern behind every loser: commodity content, bottom-dollar ad rates, or legal landmines.
Then Claude went hunting in the opposite direction: boring industries with public data, real buyers, and sponsors who happily pay to reach small, focused lists.
Seventeen niche pockets made the long list. Five survived the scoring. And then there were two: the wave of retiring baby boomer business owners selling their companies... and franchise intelligence.
Here's the fun part. On paper, the boomer exit wave scored slightly higher, because that audience is a little easier to reach. Both were winners. But I picked franchise intelligence anyway, for the most unscientific reason in this whole system: I'll ENJOY it more.
That matters more than people admit. You're going to write about your niche every single week. Pick the one you'd read about for free.
Now why is franchising so juicy?
Because the money moving through that market is stupid. A "free" franchise consultant collects a commission of 40 to 50% of the franchise fee when they place a buyer, typically $20,000 to $25,000 per signature. Brokers on business sales pull $15K to $35K per closed deal. And the buyers themselves are about to bet $150,000 to $400,000 of their savings on the biggest purchase of their life.
There's already one franchise newsletter (The Wolf of Franchises, 32,000 subscribers) supporting six figure partnership budgets inside a media company. One incumbent isn't saturation. It's proof of payroll.
And here's what Claude's research uncovered that made me grin: almost every piece of "advice" in that market is paid for by the sellers. Nobody with a big audience gets paid to tell the buyer the truth.
That's an empty seat at a rich table. We're taking it.
(One thing you'll never see us do in this niche: promise anybody earnings. The FTC just handed down its largest franchise refund order ever this March, $17 million, exactly for that kind of game. We publish documents and data, not dreams.)
The Market Disrupter Prompt
Once you have your niche, you need your angle.
This next prompt is the one I'm most protective of. I used it in 2024 to find and start AgentSkills, and that business has made almost a million dollars already.
I call it the Market Disrupter Prompt because it does three jobs in one shot. It maps every business model already making money in your market. It finds the gaps those businesses leave open. Then it hands you business models, product ideas, and marketing strategies to carve out your own slice of territory.
Remember, you only need 0.1% of a billion dollar market to make $1 million. Many markets are multiple billions. Some are trillions.
Here's the full prompt, yours to steal:
Embody a world class product developer. You have decades of experience doing extremely meticulous market research to find new hooks, angles, product ideas, feature ideas, and business models. For this task I want you to map out the business models of the [target customer] market from services to software to educational products and more. After that, I want you to research gaps in the market based on consumer demands and trends. Finally, I want you to use all of that information to create a 20 page report on ways to disrupt the [target customer] industry based on gaps, trends, unbundling, and out of the box ideas. Give me business models, product ideas, and marketing strategies that would allow for market disruption.
And here's exactly how I edited it for our new niche:
Embody a world class product developer. You have decades of experience doing extremely meticulous market research to find new hooks, angles, product ideas, feature ideas, and business models. For this task I want you to map out the business models of the franchise selling and buying market from services to software to educational products, newsletters, and more. After that, I want you to research gaps in the market based on consumer demands and trends. Finally, I want you to use all of that information to create a 20 page report on ways to disrupt the franchise selling and buying industry with an email newsletter based on gaps, trends, unbundling, and out of the box ideas. Give me business models, product ideas, and marketing strategies that would allow for market disruption.
Claude came back with a 20 page report mapping who gets paid what in franchising, ten gaps nobody is filling, eight business models, fifteen product ideas, and seven marketing strategies. You'll see the best pieces of it all month long.
Then we went one layer deeper, with research most people never do.
There's a free Claude skill called last30days (a skill is a set of instructions you install that teaches Claude a repeatable job). It sweeps Reddit, TikTok, Instagram, X, and YouTube for the last 30 days of real conversation in any market, then ranks what people are saying by actual engagement. It uses a data service called ScrapeCreators to pull all that social data. Our whole research month so far has burned about 70 of the 25,000 credits I loaded.
We pointed it at the franchise buying market and it came back with treasure.
The top influencers, with receipts on who pays them. The biggest "watchdog" YouTube channel in franchising, 181,000 subscribers, is owned by a commission brokerage. The watchdog is on the payroll.
The top subreddits and TikToks where buyers actually talk.
And most important: the pains and the gains in the buyers' own words. What they dream about, what they're terrified of, what they regret buying. Their heaven island and their hell island.
From all that, Claude built us a full empathy map (a one page picture of what your customer thinks, feels, sees, says, and does). That map is how you know which marketing messages will convert... and which ones will repel.
The Key to High Converting Offers
Everything above is market level. Money gets made at the PERSON level.
So the last step this week was turning all that research into two highly detailed customer personas. Marketers call these ICPs, ideal customer profiles: a document that describes one specific reader so vividly you could pick them out of a crowd.
They're built from actual user data scraped off the Internet, not guesses. Real Reddit threads, real reviews, real buyer surveys.
That's the secret to offers that convert: knowing the customer better than they know themselves. We know what makes them buy, what they buy, and why they buy. We know their dreams, their frustrations, the books they read, the YouTube channels they watch.
The prompt:
Embody a world class digital marketer. You have decades of experience doing meticulous market research to create highly detailed customer persona documents.
For this task I want you to use the two attached example personas as well as all the research we've done so far to create the ideal reader personas for Dana and Marcus. Make sure that the persona documents for Marcus and Dana both include all of the sections from the example personas provided. Do not use the data from the example personas for Marcus and Dana, instead use the actual research we found together for their personas (some overlaps are to be expected).
And here are the two example personas from my library that I attached as the structural template, so Claude knew exactly what sections a finished persona needs:
What came back gave me chills. Meet Dana, a 38 year old nurse seven weeks into researching her first franchise, drowning in a 212 page disclosure document. And Marcus, a 41 year old operations manager who got laid off in January, armed with a spreadsheet, a severance clock, and one question he now asks every advisor: "Who pays you, and how much?"
Those two people are who we build everything else for. The website, the opt-in page, the content, the ads. All of it.
What Happens Thursday
Add it up. An entire niche selection system ran, a 20 page disruption report got written, the voice of a whole market got mined off social media, and two complete customer personas got built.
My total hands-on time? About 20 minutes. Mostly reading and making decisions.
That's the whole point of this series. AI does the work. The human does the choosing.
Thursday in #034: we use AI to build out the website and opt-in pages for our new Zero Human Company... FranchiseFinePrint.com. Build Notes+ members get the step-by-step.
Continue with Build Notes+
This deep dive is free in full. For complete premium builds and supporting files, join Build Notes+.
