The Starting Point: What I Had Before Claude Touched Anything
Let me be clear about what I brought to this project. I had:
- 12 years of client conversations about buying property in Madeira
- A folder with 34 PDFs — my own notes, legal summaries, and market reports I’d written for clients
- A rough list of 8 topic areas I knew buyers always asked about
- Zero course-writing experience
- Zero instructional design training
What I did not have: an outline, a script, a slide deck, a sales page, or any idea how to price the thing. Claude helped me build all of that. But the underlying knowledge — the actual substance of the course — came from me. I want to be honest about that because I’ve seen people try to use AI to manufacture expertise they don’t have. It doesn’t work. Claude amplified what I already knew. It didn’t invent it.
My Real-World Experience: 23 Days From Idea to First Sale
Here’s how the 23 days actually broke down, with real time logs I kept in Notion.
Days 1–3: Building the Course Architecture
I uploaded my 34 PDFs into a Claude project and gave it a single prompt: “You are helping me design an online course for English-speaking foreigners who want to buy property in Madeira, Portugal. Based on these documents and this topic list, propose a full course structure with module names, learning objectives for each module, and a logical sequence.”
Claude returned a 11-module outline in about 40 seconds. I spent the next 3 hours editing it — moving modules around, cutting two that were too niche, combining two others. The AI gave me a skeleton I could react to, which is infinitely faster than staring at a blank document. Before this, I had spent three weekends trying to write an outline manually and getting nowhere. Those three weekends probably cost me 12 hours of real work time. Days 1 through 3 with Claude cost me maybe 4 hours total to get to a final architecture I was confident in.
Days 4–14: Writing 11 Modules of Course Content
This is where the time savings got dramatic. My process for each module:
- I wrote a 200–400 word brain dump of everything I know about the topic — messy, no structure, just getting it out
- I fed that brain dump to Claude with the module’s learning objectives and asked it to turn this into a structured lesson with an intro, three to four main sections, key takeaways, and one practical exercise
- I read the output and edited it heavily — correcting facts, adding Madeiran-specific nuance, cutting anything generic
- Claude revised based on my corrections
- I recorded the video using the script as a guide, not reading it word for word
Average time per module before this process (based on my failed attempts in 2024): roughly 3.5 hours per module. Average time per module using Claude: 55 minutes. Across 11 modules, that’s 38.5 hours reduced to just over 10 hours. I got 28 hours back in 10 days. As a solo operator still running active client consultations, that margin is the difference between a project happening and not happening.
Days 15–19: Sales Page, Email Sequence, and Pricing Research
I used Claude to draft a sales page. I gave it my course outline, my target buyer profile (English-speaking retirees and remote workers considering Madeira), and three competing courses I’d found on Teachable. It produced a full long-form sales page in one pass. I rewrote about 40% of it — mostly the sections where it got too formal and lost my voice — but the structure and the persuasion framework it used (problem, agitation, solution, proof, offer) saved me from having to figure that out from scratch.
For pricing, I asked Claude to analyze the three competitor courses and suggest a positioning strategy based on my differentiation (hyperlocal expertise, active consultant, direct Q&A access). It suggested a €297 price point for the core course and a €497 bundle with a 45-minute consultation. I went with exactly that. My first sale came in at the €497 tier.
The 5-email launch sequence — from teaser to cart-close — took Claude about 20 minutes to draft and me another 90 minutes to edit into something that felt like I actually wrote it.
Days 20–23: Setup and Launch on Teachable
I used Teachable at $39/month to host everything. Claude helped me write the course descriptions, the module intro blurbs, and the confirmation emails that go out automatically after purchase. None of that is glamorous work, but all of it would have taken me hours. With Claude, I batched all the copy for the platform in a single 2-hour session.
Launch day: I sent the sequence to my email list of 847 subscribers (people who’d contacted me over the years about Madeira property). In the first 72 hours: 6 sales — 4 at €297 and 2 at €497. Total: €2,182 from a list I’d never sold anything to before.
Exact Prompts That Did the Most Work
I’m not going to give you vague advice like “prompt it well.” Here are three prompts I used repeatedly that actually produced usable output:
For module drafting: “Here is my raw knowledge dump about [topic]. The learning objective for this module is [objective]. The student has already completed modules on [previous topics]. Write a structured lesson with a brief intro, three main sections with subheadings, a practical exercise they can do this week, and three key takeaways. Tone: direct, knowledgeable, not academic. This is a paid course for adults making a major financial decision.”
For sales copy: “Here is a description of my course and my target buyer. Write a sales page section that addresses their top three objections: [list objections]. Do not use superlatives. Do not say ‘comprehensive’ or ‘ultimate.’ Keep it factual and specific.”
For email sequence: “Write a 5-email launch sequence for a course priced at €297. Email 1 is sent 7 days before cart opens, Email 5 is cart-close day. The course is about [topic]. The list has not been sold to before. Avoid urgency tactics that feel manipulative. Focus on the cost of inaction.”
Where Claude Fell Short — Honest Limitations From This Project
I’d be doing you a disservice if I stopped at the wins. Here’s where Claude genuinely let me down during this project.
It can’t verify Portuguese legal accuracy. This is the big one for my use case. Claude would write confident, fluent paragraphs about Portuguese property law that were technically wrong — not wildly wrong, but wrong enough that a buyer acting on them could make a costly mistake. I caught several of these because I know the material. If you’re building a course on a topic where you are not the expert, Claude’s confident-sounding errors are dangerous. Every legal and regulatory claim I made had to be verified manually against the official Autoridade Tributária guidance or checked with my solicitor contact in Funchal. That added maybe 6 hours back into the project that I hadn’t budgeted for.
The voice drift problem. Across 11 modules, Claude gradually shifted tone. Early modules had my direct, slightly impatient style. By Module 9, the output was starting to sound like a McKinsey report. I had to go back and re-edit four modules that had drifted. Giving Claude a “voice reference” at the start of each session helped, but it wasn’t a perfect fix.
It generated filler when I didn’t give it enough input. Two modules where my brain dump was thin (I know these topics less deeply) produced mediocre output full of generic advice that could apply to buying property anywhere. The quality of Claude’s output is directly proportional to the quality of what you feed it. That’s not a criticism exactly — it’s just reality — but I wasn’t expecting how obvious the gap would be.
The Full Breakdown: Time, Cost, and Return
| Item | Without Claude (estimated) | With Claude (actual) |
|---|---|---|
| Course outline | ~12 hours | 4 hours |
| 11 module scripts | ~38.5 hours | ~10 hours |
| Sales page copy | ~6 hours | 2.5 hours |
| 5-email launch sequence | ~4 hours | 1.5 hours |
| Platform copy (Teachable) | ~3 hours | 2 hours |
| Total writing time | ~63.5 hours | ~20 hours |
| AI tool cost | €0 | €20 (one month Claude Pro) |
| Platform cost | $39/month Teachable | $39/month Teachable |
| Revenue (first 72 hours) | — | €2,182 |
What I’d Do Differently If I Started This in February 2026
Three things I’d change:
Start with a voice document. Before writing a single module, I’d give Claude 2,000–3,000 words of my own writing — emails, blog posts, client reports — and ask it to produce a “voice reference” summary it could use throughout the project. Fixing voice drift in editing is more expensive than preventing it upfront.
Build a fact-check step into the workflow. For every module, I’d add an explicit 20-minute review pass focused only on factual accuracy, separate from the copy-editing pass. Blending the two meant I sometimes let a loose legal claim slip through because I was focused on fixing the prose.
Launch to a smaller segment first. I sent to my full list of 847 on day one. In retrospect, I should have sent to 100–150 people I had the warmest relationships with, gathered feedback on the course experience, and fixed any problems before the main launch. The course content was solid, but the Teachable onboarding flow had a small friction issue that two early buyers messaged me about. I fixed it, but I’d rather have caught it quietly.
My Rating and Practical Summary
Claude for course creation: 4.2/5. It cut my writing time by roughly 68% and made a 23-day launch timeline realistic for a solo operator with active client work; the deduction is for the voice drift issue and the fact that domain-specific accuracy still requires serious human review.
If you’re a solopreneur sitting on expertise you’ve never productized — and most of us are — the honest answer is that Claude removes the biggest barrier, which isn’t motivation or knowledge. It’s the sheer mechanical labor of turning what you know into structured, teachable content. That part, it handles well.
What it can’t do is know your market, verify your facts, or write with your voice without significant guidance. The solopreneurs I’ve seen fail with this approach are the ones who treat Claude as a ghostwriter for content they don’t actually understand. The ones who succeed treat it as a very fast, tireless first-draft machine that still needs an expert editor — which is you.
My course on buying property in Madeira now generates revenue I don’t have to be present for. That matters a lot when you’re running a one-person operation and every hour you spend writing is an hour you’re not spending with clients. Claude Pro costs €20/month. My first launch returned €2,182 in 72 hours. The math is not complicated.
If you want to try this yourself, start with a single module — not an eleven-module course. Pick the one topic in your field that you could explain to a smart stranger in 20 minutes. Write your brain dump. Feed it to Claude. See what comes back. You’ll know within an hour whether this process fits how you think.
Ready to build your own course workflow? I put together a free prompt template pack — the exact Claude prompts I used across all 23 days — available to subscribers of this newsletter. Drop your email below and I’ll send it over.
I spent 14 years helping people buy property in Madeira before I realized I was sitting on something people outside Portugal would actually pay to learn. The island’s Golden Visa rules, the tax incentives for non-habitual residents, the quirks of buying in a place where half the land registry is still on paper — none of that existed in plain English anywhere. I had the knowledge. What I didn’t have was six months to write a course about it.
So in January 2026, I ran an experiment. I used Claude to help me write, structure, and launch a full online course — from blank page to first paying student — in 23 days. This is the exact process I followed, what worked, what embarrassed me, and what I’d do differently.
Why I Chose Claude Over Other AI Tools for This Project
I’ve been testing AI tools seriously since early 2023. I use ChatGPT for quick research pulls. I’ve run Gemini through its paces for summarizing Portuguese legal documents. But when I need to write something long, structured, and coherent — something that sounds like me and not like a press release — Claude is the only tool that consistently delivers.
The reason matters for this project specifically. A course isn’t a collection of blog posts. It has to flow. Module 3 has to build on Module 2. The tone in the intro video script has to match the tone in the final assessment questions. I needed an AI that could hold a long context window and keep the voice consistent across 11 modules. Claude’s 200K token context window on the Pro plan (€20/month at the time I ran this) made that possible in a way ChatGPT simply couldn’t match for my workflow.
The Starting Point: What I Had Before Claude Touched Anything
Let me be clear about what I brought to this project. I had:
- 12 years of client conversations about buying property in Madeira
- A folder with 34 PDFs — my own notes, legal summaries, and market reports I’d written for clients
- A rough list of 8 topic areas I knew buyers always asked about
- Zero course-writing experience
- Zero instructional design training
What I did not have: an outline, a script, a slide deck, a sales page, or any idea how to price the thing. Claude helped me build all of that. But the underlying knowledge — the actual substance of the course — came from me. I want to be honest about that because I’ve seen people try to use AI to manufacture expertise they don’t have. It doesn’t work. Claude amplified what I already knew. It didn’t invent it.
My Real-World Experience: 23 Days From Idea to First Sale
Here’s how the 23 days actually broke down, with real time logs I kept in Notion.
Days 1–3: Building the Course Architecture
I uploaded my 34 PDFs into a Claude project and gave it a single prompt: “You are helping me design an online course for English-speaking foreigners who want to buy property in Madeira, Portugal. Based on these documents and this topic list, propose a full course structure with module names, learning objectives for each module, and a logical sequence.”
Claude returned a 11-module outline in about 40 seconds. I spent the next 3 hours editing it — moving modules around, cutting two that were too niche, combining two others. The AI gave me a skeleton I could react to, which is infinitely faster than staring at a blank document. Before this, I had spent three weekends trying to write an outline manually and getting nowhere. Those three weekends probably cost me 12 hours of real work time. Days 1 through 3 with Claude cost me maybe 4 hours total to get to a final architecture I was confident in.
Days 4–14: Writing 11 Modules of Course Content
This is where the time savings got dramatic. My process for each module:
- I wrote a 200–400 word brain dump of everything I know about the topic — messy, no structure, just getting it out
- I fed that brain dump to Claude with the module’s learning objectives and asked it to turn this into a structured lesson with an intro, three to four main sections, key takeaways, and one practical exercise
- I read the output and edited it heavily — correcting facts, adding Madeiran-specific nuance, cutting anything generic
- Claude revised based on my corrections
- I recorded the video using the script as a guide, not reading it word for word
Average time per module before this process (based on my failed attempts in 2024): roughly 3.5 hours per module. Average time per module using Claude: 55 minutes. Across 11 modules, that’s 38.5 hours reduced to just over 10 hours. I got 28 hours back in 10 days. As a solo operator still running active client consultations, that margin is the difference between a project happening and not happening.
Days 15–19: Sales Page, Email Sequence, and Pricing Research
I used Claude to draft a sales page. I gave it my course outline, my target buyer profile (English-speaking retirees and remote workers considering Madeira), and three competing courses I’d found on Teachable. It produced a full long-form sales page in one pass. I rewrote about 40% of it — mostly the sections where it got too formal and lost my voice — but the structure and the persuasion framework it used (problem, agitation, solution, proof, offer) saved me from having to figure that out from scratch.
For pricing, I asked Claude to analyze the three competitor courses and suggest a positioning strategy based on my differentiation (hyperlocal expertise, active consultant, direct Q&A access). It suggested a €297 price point for the core course and a €497 bundle with a 45-minute consultation. I went with exactly that. My first sale came in at the €497 tier.
The 5-email launch sequence — from teaser to cart-close — took Claude about 20 minutes to draft and me another 90 minutes to edit into something that felt like I actually wrote it.
Days 20–23: Setup and Launch on Teachable
I used Teachable at $39/month to host everything. Claude helped me write the course descriptions, the module intro blurbs, and the confirmation emails that go out automatically after purchase. None of that is glamorous work, but all of it would have taken me hours. With Claude, I batched all the copy for the platform in a single 2-hour session.
Launch day: I sent the sequence to my email list of 847 subscribers (people who’d contacted me over the years about Madeira property). In the first 72 hours: 6 sales — 4 at €297 and 2 at €497. Total: €2,182 from a list I’d never sold anything to before.
Exact Prompts That Did the Most Work
I’m not going to give you vague advice like “prompt it well.” Here are three prompts I used repeatedly that actually produced usable output:
For module drafting: “Here is my raw knowledge dump about [topic]. The learning objective for this module is [objective]. The student has already completed modules on [previous topics]. Write a structured lesson with a brief intro, three main sections with subheadings, a practical exercise they can do this week, and three key takeaways. Tone: direct, knowledgeable, not academic. This is a paid course for adults making a major financial decision.”
For sales copy: “Here is a description of my course and my target buyer. Write a sales page section that addresses their top three objections: [list objections]. Do not use superlatives. Do not say ‘comprehensive’ or ‘ultimate.’ Keep it factual and specific.”
For email sequence: “Write a 5-email launch sequence for a course priced at €297. Email 1 is sent 7 days before cart opens, Email 5 is cart-close day. The course is about [topic]. The list has not been sold to before. Avoid urgency tactics that feel manipulative. Focus on the cost of inaction.”
Where Claude Fell Short — Honest Limitations From This Project
I’d be doing you a disservice if I stopped at the wins. Here’s where Claude genuinely let me down during this project.
It can’t verify Portuguese legal accuracy. This is the big one for my use case. Claude would write confident, fluent paragraphs about Portuguese property law that were technically wrong — not wildly wrong, but wrong enough that a buyer acting on them could make a costly mistake. I caught several of these because I know the material. If you’re building a course on a topic where you are not the expert, Claude’s confident-sounding errors are dangerous. Every legal and regulatory claim I made had to be verified manually against the official Autoridade Tributária guidance or checked with my solicitor contact in Funchal. That added maybe 6 hours back into the project that I hadn’t budgeted for.
The voice drift problem. Across 11 modules, Claude gradually shifted tone. Early modules had my direct, slightly impatient style. By Module 9, the output was starting to sound like a McKinsey report. I had to go back and re-edit four modules that had drifted. Giving Claude a “voice reference” at the start of each session helped, but it wasn’t a perfect fix.
It generated filler when I didn’t give it enough input. Two modules where my brain dump was thin (I know these topics less deeply) produced mediocre output full of generic advice that could apply to buying property anywhere. The quality of Claude’s output is directly proportional to the quality of what you feed it. That’s not a criticism exactly — it’s just reality — but I wasn’t expecting how obvious the gap would be.
The Full Breakdown: Time, Cost, and Return
| Item | Without Claude (estimated) | With Claude (actual) |
|---|---|---|
| Course outline | ~12 hours | 4 hours |
| 11 module scripts | ~38.5 hours | ~10 hours |
| Sales page copy | ~6 hours | 2.5 hours |
| 5-email launch sequence | ~4 hours | 1.5 hours |
| Platform copy (Teachable) | ~3 hours | 2 hours |
| Total writing time | ~63.5 hours | ~20 hours |
| AI tool cost | €0 | €20 (one month Claude Pro) |
| Platform cost | $39/month Teachable | $39/month Teachable |
| Revenue (first 72 hours) | — | €2,182 |
What I’d Do Differently If I Started This in February 2026
Three things I’d change:
Start with a voice document. Before writing a single module, I’d give Claude 2,000–3,000 words of my own writing — emails, blog posts, client reports — and ask it to produce a “voice reference” summary it could use throughout the project. Fixing voice drift in editing is more expensive than preventing it upfront.
Build a fact-check step into the workflow. For every module, I’d add an explicit 20-minute review pass focused only on factual accuracy, separate from the copy-editing pass. Blending the two meant I sometimes let a loose legal claim slip through because I was focused on fixing the prose.
Launch to a smaller segment first. I sent to my full list of 847 on day one. In retrospect, I should have sent to 100–150 people I had the warmest relationships with, gathered feedback on the course experience, and fixed any problems before the main launch. The course content was solid, but the Teachable onboarding flow had a small friction issue that two early buyers messaged me about. I fixed it, but I’d rather have caught it quietly.
My Rating and Practical Summary
Claude for course creation: 4.2/5. It cut my writing time by roughly 68% and made a 23-day launch timeline realistic for a solo operator with active client work; the deduction is for the voice drift issue and the fact that domain-specific accuracy still requires serious human review.
If you’re a solopreneur sitting on expertise you’ve never productized — and most of us are — the honest answer is that Claude removes the biggest barrier, which isn’t motivation or knowledge. It’s the sheer mechanical labor of turning what you know into structured, teachable content. That part, it handles well.
What it can’t do is know your market, verify your facts, or write with your voice without significant guidance. The solopreneurs I’ve seen fail with this approach are the ones who treat Claude as a ghostwriter for content they don’t actually understand. The ones who succeed treat it as a very fast, tireless first-draft machine that still needs an expert editor — which is you.
My course on buying property in Madeira now generates revenue I don’t have to be present for. That matters a lot when you’re running a one-person operation and every hour you spend writing is an hour you’re not spending with clients. Claude Pro costs €20/month. My first launch returned €2,182 in 72 hours. The math is not complicated.
If you want to try this yourself, start with a single module — not an eleven-module course. Pick the one topic in your field that you could explain to a smart stranger in 20 minutes. Write your brain dump. Feed it to Claude. See what comes back. You’ll know within an hour whether this process fits how you think.
Ready to build your own course workflow? I put together a free prompt template pack — the exact Claude prompts I used across all 23 days — available to subscribers of this newsletter. Drop your email below and I’ll send it over.
Robson Penassi
Real estate consultant in Madeira, Portugal. Solopreneur since 2012. Testing AI tools since 2023 to automate his one-person business. Writes about what actually works — and what does not.
More articles by Robson →