You Bring the Expertise. AI Brings the Speed.

Yes. You can use AI to build your course, and I'd tell you to. I didn't have any of this when I built the Elite Nurse Practitioner. I still write most of what I make myself. And I use a chat window every day now, for exactly the parts of the job it's good at.

Here's the catch. Most people use it in the one spot where it can't help them. They open a window, type "write me a course on wound care," read three paragraphs that could have come off any website, and decide one of two things. Either the machine can do this for them, so they don't have to. Or it's useless, so they close the tab. Both of those people have a deck with zero slides in it a month later.

Use it before you build and after you build. The building itself — the part where what you know goes onto the slides — that's you. Find where you are below and start there.

Door one: you don't have the idea yet

If you already know what you're building, skip this door. Don't waste your time here.

If you don't, this is the part I'm jealous of, because I didn't have it. Open a chat window and tell it who you are — profession, credentials, years, specialty, the things colleagues actually ask you about — and then ask it to identify twenty areas where you have expertise that other professionals might pay to learn in a course. Clinical, operational, business, implementation, all of it. The questions behind that prompt are the ones I'd have you ask yourself anyway; the window just asks them faster and doesn't let you off with "nothing."

Then push it further. Ask for twenty-five course ideas that solve real problems for people in your profession, and for each one, the audience, the problem, the outcome, and the revenue potential. And when something on that list makes you sit up, run the market check: analyze the opportunity for a continuing education course on this — demand, competitors, who buys it, where it's growing, the risks — and give me a recommendation. It'll go read the subreddits and the Facebook groups and come back in a few minutes with what it found. That used to be a marketing agency's job, and they charged thousands of dollars for it.

Some rules of thumb. The more you put in, the better what comes out; a two-line prompt gets a two-line answer. Some of what it gives you is garbage. It's right most of the time and wrong a lot of the time. So don't trust the first run. Run it three or four times with different inputs and watch what keeps showing up — the profitable ideas are the ones that appear across runs.

And one honest check. Most of the course ideas I ever had — I'd say seventy-five percent — came from sitting in groups listening to what people were complaining about. The window is a faster way to do that listening; it doesn't replace it.

Now the trap, which I put in my own course on purpose. Research is free now and there's no end to it, so it's a very comfortable place to sit and not start. If the market check says there's demand and a little looking of your own agrees, that's your answer. You are not going to get a better one by asking again. Stop going to AI. Declare the topic and walk through door two.

Door two: you have the idea and a blank deck

Before anything else, make a project. Every major chat tool has them. Name it after your business and tell it, once, who you help, what problem you solve, and what outcome you promise. Upload anything you've already got. Everything you tell it or give it stays in that project, so every prompt after that starts from what it already knows about your course instead of from zero.

Then the prompt that does the real work. Tell it you want to build a presentation for these people about this problem, that the solution you teach runs like this — and give it your actual steps — and ask it to generate an outline of about a hundred slides. Then add one line: "Do you need any additional parameters from me?" It will. Answer them. What you get back is an outline — the same outline anyone with your prompt would get. I've written about why that's worth nothing on its own. It's worth a great deal as a place to start.

Now the middle. You build the slides. The content on them — what you actually do, the step everyone gets wrong, the thing the textbook says that quietly stops being true in practice — comes out of your head and nowhere else. This is where the hours go, and they're yours to spend. How I build a deck and record it is its own article.

Two things the window does well inside that middle. When you're stuck on a slide, paste the slide's title in and ask it to help you hash out the details. It gets the juices flowing again. And when a slide is done, paste the content in and ask it to review for errors and add anything relevant you've missed. Then you decide what stays, because you're the one who knows.

A word on the tools that promise to skip the middle. Slide generators — Google has one, Chat Slide is another — take your outline and hand you back a deck. What I've gotten out of them is generic; there's no instinct in it for what matters to someone doing the job. Use one to lay out a template if you like, then get in and put the content in yourself. A friend of mine built ContentCow for exactly this job — outline, lessons, the course, the marketing content around it — and it'll take you from outline to slides in about four hours. You still have to record it.

Images: don't pull them off the internet. I did that once, a photo of a guy lifting weights, the most generic picture in the world, and I got a cease-and-desist letter in the mail. Use the stock images built into PowerPoint, or have the image tool inside ChatGPT — or Canva, or Midjourney — make you what you need. Same for charts: describe the chart and it draws it, and then you check the numbers on it the way you'd check a slide.

Door three: you'd rather talk than type

Some people can't type a slide to save their lives but can explain the whole process to a new hire over lunch. If that's you, talk instead. Tools like Otter or Descript will record you and turn what you said into lecture and slide content. Then you edit it, because it will mishear you in places, and you build the deck from what's left. Same middle, different way in.

After the deck

Upload the finished deck into your project and ask for the handouts — a one-page PDF of the process from module three, a checklist for the thing in module five. It won't be perfect. Don't download it and throw it into the course. Open it, edit it, make sure it isn't regurgitated nonsense.

The boring documents are a good use of it too. Make a subfolder in the project for legal, give it a template, tell it the phrases that have to be in there — for informational purposes only, your refund policy, your money-back window — and have it build out a disclaimer and a terms of service. Then read every line. And the more liability there is in what you teach — medical education that could hurt a patient, tax advice that could get somebody in trouble — the more robust that document has to be, and that's the one place I'd have a lawyer read what the machine wrote before it goes on your site.

One fence. The articles that market your course are a different job with a different rule — I've seen AI-written articles come back plagiarized from another site — so read that before you carry this workflow over there.

The line it can't cross

Everything above assumes one thing: that you know the material. If you don't — if it's something you've read about rather than done — an AI-built course is you making it up, and your students will be the ones who find out. A window can produce a perfectly competent generic course all by itself. If you want to pump out generic CE and sell it cheap, by all means; it works, it can be a profitable business. It just isn't the one that gets built around your name.

So the rule is the one I give people in my own course: you bring your expertise, AI helps bring the speed. I used it sparingly building that course — some ideas, organizing, formatting. Everything in it came out of my head, and that's the part anyone pays for.

Tonight

Open a chat window — ChatGPT, Claude, Gemini, whichever you'll actually use. Pay for the better tier; it's ten or twenty dollars a month. Make a project and name it after your business. Tell it who you help, what problem, what outcome.

If you have the idea: paste in the problem and the solution and ask for the hundred-slide outline. Answer its questions. Save the outline.

If you don't: run the twenty-areas prompt three times and circle what repeats.

Either way, you'll have a document by bedtime that you didn't have this morning. Tomorrow, start on the slides.