How I Built a Complete Course Using Claude in 2 Weeks

I built a full online course in 11 days. No team. No course creation agency. No $3,000 ghostwriter. Just me, my laptop, and Claude — and I want to be completely honest about how that actually went, including the parts that frustrated me.

For context: I run a solo real estate consulting business in Madeira, Portugal. I have been doing this since 2012. In 2023 I started testing AI tools seriously, not as a hobby, but because I needed to scale output without hiring. By early 2026, I had a waiting list of expats wanting to buy property in Madeira, and I kept getting the same 40 questions on every discovery call. I realized I was essentially teaching the same intro course every single week for free. That had to stop.

So I decided to build a paid course: Buying Property in Madeira: What Expats Need to Know Before They Sign Anything. Here is exactly what I did, what worked, what didn’t, and whether another freelancer should try this approach.

Why I Chose Claude Over Other AI Tools for This Project

I have ChatGPT, Gemini, and Claude all running on paid plans. I use them for different things. For long-form structured writing — anything that needs to stay internally consistent across thousands of words — Claude wins for me, consistently. It holds context better in long conversations. When I am writing a 2,000-word module and refer back to something I said in the outline three messages ago, Claude usually tracks it. ChatGPT drifts. That matters enormously when writing a course where Module 4 has to logically follow Module 2.

I was on Claude Pro at $20/month at the time. That is the only tool cost I am counting here because I already owned Notion (where I organized everything) and Loom (where I recorded the videos).

The Problem I Was Trying to Solve as a Solo Operator

Before I started this project, I tracked my time for one month. I was spending roughly 6 hours per week on introductory consultations with leads who were not yet ready to buy. These were people who needed education, not a consultant. They needed a course. I was giving it away for free, one call at a time.

Creating a course from scratch felt overwhelming. I had tried twice before in 2023 and abandoned it both times before finishing the outline. The content wasn’t the issue — I know this material cold. The bottleneck was structuring it, writing it, and producing it fast enough that I didn’t lose momentum.

That’s the exact problem Claude solved. Not the expertise. The execution.

My Exact Process: How I Used Claude to Write the Entire Course

Step 1: Brain Dump Into a Master Prompt

I opened a new Claude conversation and spent about 45 minutes doing a raw brain dump. I typed everything I normally cover in those intro calls: legal structures for foreign buyers, the NHR tax regime changes, the Golden Visa situation, due diligence steps, common traps with off-plan purchases, how promissory contracts work in Portugal, notary fees, IMT tax calculations, finding a reliable lawyer. Everything.

I then asked Claude to organize this into a course structure with modules, lessons per module, and a one-sentence description of each lesson. I told it the audience was English-speaking expats, mostly from the UK, US, and northern Europe, with no prior knowledge of Portuguese property law.

It produced a clean 6-module structure in about 90 seconds. I spent another 30 minutes revising it — moving two lessons, cutting one module down, adding a bonus section on hiring a buyer’s agent. That outline became my bible for the next 11 days.

Step 2: Writing Each Lesson Script

For each lesson, I gave Claude the lesson title, the module context, the target audience definition I had written, and three to five bullet points of what I wanted to cover. I also pasted in any notes from my brain dump that were relevant. Then I asked for a conversational script — not an essay, not bullet points, but something I could read to camera with minor ad-libbing.

The first draft of each script came back in about 600-900 words. I would read it, make corrections where the Portuguese legal details were slightly off (Claude is good but it does not know the specific 2026 IMT rates by heart — more on this below), and then ask it to tighten the opening and sharpen the summary at the end.

Total lessons: 34. Average time per lesson from my side: about 18 minutes. That includes reading the draft, correcting it, asking for revisions, and finalizing. Old process — writing scripts manually — took me about 45 minutes per lesson on the ones I had completed in 2023. The math: 34 lessons × 27 minutes saved = roughly 15 hours saved on script writing alone.

Step 3: Workbook and Supplementary Materials

After the scripts, I used Claude to write the student workbook exercises — one reflection question and one action step per lesson. I also had it draft a glossary of Portuguese real estate terms (30 terms), a buyer’s checklist, and a “red flags” reference sheet for reviewing promissory contracts. These materials would have taken me an entire weekend to produce manually. Claude produced rough drafts of all of them in one afternoon session of about 3 hours, and I spent another 2 hours editing for accuracy.

Step 4: Sales Page and Email Sequence

I also used Claude to write the sales page copy and a 5-email launch sequence. The sales page took two rounds of prompting and about 40 minutes of editing on my end. The emails took one session of about 90 minutes total, including my revisions. I have run these tools enough to know that sales copy from Claude needs a human pass for tone — it can sound slightly generic if you do not push it with specifics about your audience and your own voice.

My Real-World Experience: Madeira Real Estate Consulting, January 2026

Let me give you the specific picture. I launched the course on January 28, 2026 at €297. In the first two weeks, 14 people bought it — €4,158 in revenue from a product that cost me €20 (one month of Claude Pro) and approximately 38 hours of my time across 11 days. That 38 hours includes everything: the brain dump, all 34 lesson scripts, the workbook, the sales page, the email sequence, recording setup, and uploading to Teachable.

For comparison: in 2023, when I tried to build this course manually, I logged 22 hours over six weeks and produced 9 lesson scripts before abandoning it. I had nothing to sell. The difference was not intelligence or effort — it was momentum. Claude kept me moving. Every time I sat down, I had something concrete to work with within minutes instead of staring at a blank document trying to remember how I wanted to sequence the NHR tax content.

The module I am most proud of is Module 3: Due Diligence Before You Sign. It covers caderneta predial (property register), licença de habitação (habitation license), and how to check for illegal constructions — things that trip up foreign buyers constantly in Madeira. The script Claude produced for that module was surprisingly solid on structure. I had to correct two specific details (a fee amount and a reference to a regulation that had been updated in 2026), but the teaching flow, the analogies it used for a non-Portuguese audience, and the pacing were genuinely good. Better than my first draft would have been.

I also used Claude to write the course FAQ page, which I had completely forgotten to plan for. I described my typical buyer personas — “the remote worker from the UK considering NHR,” “the retiree couple from Germany looking for a holiday home,” “the American investor interested in rental yield” — and asked Claude to generate 12 likely questions and draft answers. I edited the answers for accuracy and published it. Took 45 minutes. That FAQ page has reduced my post-purchase support emails noticeably in the first month.

The one thing I did differently from what I’d normally do: I treated Claude as a first-draft engine, not a final-draft machine. Every single output went through my eyes and my real estate knowledge before it went to a student. That filter matters. Without it, you will publish mistakes.

What Claude Does NOT Do Well in This Kind of Project

I want to be direct about this because the failures are just as useful as the wins.

It gets jurisdiction-specific legal and tax details wrong. Portuguese real estate law is specific. IMT rates, AIMI thresholds, NHR eligibility rules — these change. Claude’s training data has a cutoff and it does not know about regulatory updates from the past year. In three lessons, I caught specific numbers that were outdated or slightly off. If I had published those without checking, I would have given students incorrect information about real money decisions. This is not a minor flaw. For anyone using Claude to write content about law, tax, finance, or medicine — you must verify every specific claim against current sources. Every single one.

The sales copy defaults to safe and slightly bland. Claude writes technically competent sales copy but it tends toward the middle. The emotional edge — the “I know exactly what you’re afraid of and here is why this course addresses it” — needed significant rewriting from me. I pushed it with more specific prompts about buyer fears and it improved, but the first pass was not something I would have published without heavy editing.

Long sessions can produce repetition across lessons. By the time I was writing Module 5, I noticed Claude occasionally recycling phrasing from earlier modules. I had to specifically prompt: “Check this lesson for any phrasing that overlaps with what we covered in Module 2 and vary the language.” That worked, but it required active management.

Time and Cost Breakdown: What This Project Actually Cost Me

Task	Time With Claude	Estimated Time Without Claude	Time Saved
Course outline	1.5 hours	4 hours	2.5 hours
34 lesson scripts	10 hours	25.5 hours	15.5 hours
Workbook + materials	5 hours	12 hours	7 hours
Sales page + email sequence	3.5 hours	8 hours	4.5 hours
FAQ page	0.75 hours	3 hours	2.25 hours
Total	~20.75 hours	~52.5 hours	~31.75 hours

Tool cost: €20 for one month of Claude Pro. Recording and hosting (Loom + Teachable) I already paid for. Net result in first two weeks: €4,158 revenue on €20 in new tool spend.

What I Would Do Differently If I Built Another Course Today

Start with a “persona document” in the very first message. I did this informally, but if I do this again, I will write a proper 300-word description of my target student — their fears, their vocabulary, their prior knowledge level, what they have already googled — and paste that at the top of every new Claude session. It would have saved me time re-explaining context each day.

I would also create a separate Claude conversation just for fact-checking. Instead of mixing writing and verification in the same thread, I would paste each finished script into a clean conversation and ask Claude to flag any claims that depend on specific numbers, dates, or regulations so I know exactly what to verify manually. That’s a smarter workflow than trying to remember what to check after the fact.

And I would record first, script second on at least some lessons. A few of my best lessons ended up being ones where I talked through the concept on Loom first, then had Claude clean up the transcript into a polished script. The spontaneity came through better.

Should You Try This as a Freelancer?

If you are a freelancer or solo consultant sitting on expertise that you keep giving away for free — on calls, in emails, in repeated explanations — yes. Build the course. Use Claude. The bottleneck was never your knowledge. It was the time and friction to structure and write it.

Claude handles structure well. It handles first drafts well. It holds long context better than the other tools I have tested. What it does not do is replace your domain expertise or your quality filter. You have to know your subject well enough to catch its mistakes — and it will make mistakes, especially on anything jurisdiction-specific, regulatory, or time-sensitive.

My rating: 4.5/5 — specifically for long-form, multi-part course creation, because the context retention across a 34-lesson project is genuinely better than anything else I have used, and it cut my production time nearly in half.

Practical Summary Before You Start

• Do a full brain dump in your first Claude session before asking for any structure
• Get the outline locked before writing a single lesson — changes to structure mid-project cost more time than they save
• Write a detailed audience persona document and use it in every session
• Verify every specific number, fee, regulation, or legal claim against current sources before publishing
• Edit sales copy harder than you edit lesson scripts — Claude’s default tone is safe and needs sharpening
• Claude Pro at $20/month is all you need for this scale of project

If you have been sitting on a course idea for more than six months, that is six months of potential revenue you have not collected. I left it on the table for two years. The tools are good enough now that the only thing stopping you is starting.

Start with your brain dump today. Open Claude, describe your expertise and your ideal student, and ask it to help you structure what you already know. The outline alone will show you whether this is viable — and it will take you less than two hours to find out.