๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you๐Ÿค– Build Your Own AI Agent โ€” 10 modules, from zero to a 24/7 AI employee working for you
Back to course
Module 2 ยท ~8 minutes

Module 2: Choosing Your Course Topic

AI Course Creation

READ
David Perell teaches an online writing course called "Write of Passage." It costs $4,000. He could have taught "creative writing" โ€” a topic with 100x more search volume. Instead, he chose "online writing for professionals" โ€” a narrow, specific, high-value topic. His course has generated over $10 million.

The lesson: popular topics are bad course topics. The more popular the topic, the more competition you face, the lower you can charge, and the harder it is to stand out. The riches are in the niches.

The Topic Selection Framework

A great course topic sits at the intersection of three things:

1. Your Expertise (What you know)
Not necessarily formal credentials โ€” lived experience counts. Have you done this thing? Have you helped others do it? Can you get someone from A to B reliably?

Red flag: If you'd need to research extensively just to create the content, you don't know enough yet. Teach what you've done, not what you've read about.

2. Market Demand (What people pay for)
Test demand with real signals, not assumptions:

  • Udemy: courses exist with 10,000+ students

  • Google: people search for "[topic] course" or "how to [topic]"

  • Reddit/communities: people ask about this regularly

  • Competitors: others charge ยฃ200+ for courses on this topic


No demand signals = no market. Don't create a course nobody's searching for.

3. Transformation Value (What it's worth)
The best courses deliver a transformation โ€” a measurable before/after. The more valuable the transformation, the more you can charge.

Quick Check

Which prompt creates a better course curriculum?

Which prompt is better?

Low value: "Learn the basics of photography" (ยฃ30 course)
High value: "Build a photography portfolio that lands paid clients" (ยฃ500 course)

The transformation should be: specific, measurable, and desirable. "Learn AI" is vague. "Use AI to earn your first ยฃ1,000 as a freelancer" is specific, measurable, and desirable.

The Niche-Down Method

Start broad, then narrow three times:

  • Level 1: "Marketing" โ†’ too broad, infinite competition

  • Level 2: "Email marketing" โ†’ still broad but more focused

  • Level 3: "Email marketing for e-commerce brands" โ†’ good, specific audience

  • Level 4: "Email marketing automations for Shopify stores" โ†’ excellent, very specific, high-value audience


At level 4, you have a clear audience (Shopify store owners), a specific skill (email automations), and high willingness to pay (e-commerce businesses invest in marketing). Your course practically sells itself.
Quick Check

What makes online course content stick with students?

Topics to Avoid

Anything you can't demonstrate results for. If you can't show before/after outcomes, students will demand refunds.

Anything free on YouTube. If comprehensive, free versions exist, your course needs to offer structure, community, or accountability to justify the price. Raw information isn't enough.

Anything that changes faster than you can update. A course on "the latest AI tools" will be outdated in 3 months. A course on "AI workflows for [specific outcome]" has longer shelf life because principles persist even when tools change.

Anything you're not passionate enough to talk about for 6 months. You'll need to create content, answer questions, update materials, and market the course. If the topic bores you, you'll abandon it.

Using AI to Validate and Refine

Quick Check

Course curriculum should be designed ___, starting from the end goal and working back to determine what students need to learn.

Course curriculum should be designed , starting from the end goal and working back to determine what students need to learn.
AI is genuinely powerful for topic validation because it can synthesise information from multiple sources quickly. Use it to:

  • Analyse competitor course reviews (what students love/hate)

  • Identify underserved sub-niches within a broad topic

  • Generate potential course angles and test which resonates

  • Research the audience's language and pain points


But don't let AI make the decision for you. Your lived experience and market instinct matter more than any analysis.

---

TRY IT

Topic Validation Deep Dive

I'm considering creating a course about [TOPIC] for [TARGET AUDIENCE]. Do a comprehensive validation: (1) List the top 10 existing courses on this topic across Udemy, Skillshare, and Teachable โ€” note their prices, ratings, and student counts, (2) Identify the top 3 underserved angles based on common complaints in reviews, (3) Suggest 5 niche-down variations of this topic that would face less competition but maintain strong demand, (4) For each variation, rate the combination of demand, competition, and price potential (1-10). Recommend the strongest option with reasoning.

Transformation Statement Creator

I want to create a course about [TOPIC]. Help me craft a powerful transformation statement. Generate 10 options using this formula: "Go from [SPECIFIC BEFORE STATE] to [SPECIFIC AFTER STATE] in [TIMEFRAME]." Each should be: specific (not vague), measurable (the student can verify they achieved it), desirable (they'd pay money for this), and believable (not overpromising). Rank them by how compelling they'd be on a sales page. For the top 3, explain why they work and how I could prove the transformation with student testimonials.

Competitor Gap Analysis

Analyse the top 5 courses about [TOPIC] on Udemy. For each course: summarise the curriculum, identify what students praise in reviews, identify what students complain about, and note what's missing. Then synthesise: what are the 3 biggest gaps that a new course could fill? For each gap, suggest how I could specifically address it in my curriculum. Present this as a competitive positioning map showing where my course would stand vs. existing options.
EXERCISE
The Topic Shortlist

1. Brainstorm 10 topics you could teach (set a timer for 10 minutes โ€” no filtering)
2. Score each on: your expertise (1-5), market demand (1-5), transformation value (1-5)
3. Eliminate anything scoring below 3 on any dimension
4. For the top 3, niche down at least two levels using the Niche-Down Method
5. For your #1 pick, write the transformation statement: "By the end of this course, you'll be able to [SPECIFIC OUTCOME]"
6. Share this statement with 5 people in your target audience and gauge their reaction

If 3+ people say "I'd buy that," you have your topic.

---