Lesson 1 · 16 min
Selling AI is not one single thing. It comes in layers: some become more valuable over time, others turn into a commodity and the price collapses. Building a website, a store or a bot has become easy, and anyone who sells only that ends up competing on price. This lesson separates the layers and builds your portfolio on the base that lasts.
Start the lesson
Goal: separate the build offers, which lose value, from the trust offers, which accumulate, and put together your portfolio of 4 offers.
From top to base: the ticket goes down, the useful life goes up. Click on each layer.
Your identity stays at the base: "I help companies trust what AI does," not "I build voice agents."
Four reasons a company pays someone to run AI. They are not equally good.
The client pays a lot because they don't know how simple it is.
It disappears as soon as the topic becomes a popular tutorial. Use it as cash flow, never as a plan.
You teach or diagnose something small for free, the person sees real value and accepts a trial project.
It is the most reliable route for someone starting from zero.
A well-done free diagnosis opens a door that no email ever will.
Did you work for years in logistics, accounting or insurance? Sell there.
You already speak the language, know the rules and know where it hurts.
Large companies get asked internally: "what are we doing about AI?"
That's where training and consulting reach the highest fees, but a beginner rarely gets there in the first year.
The number that shows up in screenshots is always revenue. On its own, it tells you nothing.
One big contract that pays a team can leave almost nothing. A smaller project done by you can leave much more.
A healthy benchmark for an operation with subcontractors: net margin between 35% and 45%.
Count the hours spent on proposals and meetings that didn't close.
A project that's approved but not paid is a loss with a deadline.
The audit uncovers the problem; fixing that problem is the next sale.
People who have learned to use it ask for training; people who have something running ask for ongoing support.
Monthly advisory usually lasts two to three months. The antidote is in lesson 8: show the month 5 map in month 2.
Over-engineering, building on rotten data, promising a deadline that depends on third parties, picking a niche without data, and charging while apologizing.
Speak more simply than seems necessary.
The one-line calculation before you accept any project, and the portfolio prompt.
PROJETO: <nome> Preço: <valor> · Horas estimadas: <h> · Custo de terceiros: <valor> Margem = (preço − custo de terceiros − horas × <sua hora>) ÷ preço = <%> Se a margem não fecha em pelo menos um terço: [ ] o preço sobe [ ] o escopo encolhe [ ] eu recuso
You are my offer strategist. I'm going to build an AI services business for the next 90 days, selling trust: diagnosing, reviewing, supporting and teaching. Building only comes in as part of one of those services. My context: - Industry where I already have real experience: <setor, ou "nenhum"> - What I already know how to build today: <liste, mesmo que pouco> - Hours per week available: <número> - How much I need to bill per month for this to be worth it: <valor> Build for me: 1. FOUR offers, from the easiest to sell to the hardest. For each one: name, what's included, what's left out, delivery time, price range and why a client would buy THIS one. 2. The SEQUENCE: which offer opens the door to which, and the sentence I use to propose the next one after delivering the previous one. 3. The MATH: how many sales of each offer I need per month, with the hours I have. If it doesn't add up, say so plainly and propose what to cut. 4. Mark which offers will probably be commoditized in 12 months. Don't invent demand. If information is missing, ask before guessing.
Save your answers in a file with today's date. Read them again in 90 days.
Run the portfolio prompt with your context. Rewrite the result until every offer is a diagnosis, review, ongoing support or teaching; if there is building, it shows up inside one of those services.
Done when: the file has 4 offers with what's left out of each one, the sequence between them and the math closed in numbers; no offer is just "I build."
Take your entry offer and fill in the one-line calculation with price, hours and third-party cost.
Done when: the margin is calculated in % and, if it came out below one third, you marked what changes: price, scope or decline.
4 questions. The explanation appears after your attempt.
Question 1 of 4
Which layer of the pyramid becomes more valuable the longer you stay in it?
Question 2 of 4
You're starting from zero, with no background in any industry. Which door do you go in through?
Question 3 of 4
A big monthly advisory contract and a small project you did alone. Which number decides which one is better?
Question 4 of 4
Which offer in this portfolio is most likely to lose value within 12 months?
Pick one option. The explanation shows up after you try.