How to Build World-Class Courses with AI

Create engaging, high-impact, custom courses faster—using simple, powerful AI prompts.

Hi All

I'm excited to send my first real email from Substack. To celebrate, I'm sending you one of my favorite prompts to create bespoke courses tailored to your needs.

If you prefer a video walkthrough, here you are:

I first started experimenting with this prompt when I was trying to elevate my own prompting game. I watched YouTube videos, read guides, and received prompts from X. All the content was great, but it wasn't tailored to my specific needs.

I wanted to delve into how AI can be applied to marketing and go-to-market strategies, and I didn't feel that anyone had done a great job of creating advanced content in that space.

So, I turned to my favorite co-worker - o3.

I worked with one of my CustomGPTs (trained on how to create prompts for GPT -3 models), and we came up with the prompt below (after a lot of iteration).

Let me give you an example of how good the output is, with some quick call-outs on how I designed the prompt.

1. Topic: I asked for a course on ABM for SMBs. I'll discuss this further in future posts. The value AI creates for marketing lies in its ability to target micro-audiences effectively. ABM isn't something I've been a massive fan of. However, as marketing shifts from segment-level to company-level marketing and AI becomes more prevalent, account-based marketing (ABM) will become even more impactful.

This was a real test for AI because it's not a topic on which anyone has created content to date, so it would have to construct most of the lessons itself.

The prompt will create the course on any topic you want.

2. Format: I wanted all action no fluff, so you'll see in the output it's incredibly actionable. The course is designed to give me a weekly theme and then 5 daily lessons on that theme.

Example Output from the course I wanted ….

---

Week 1 — AI-Powered ICP & Micro-Segment Discovery

Strategic function: Uncover overlooked, ready-to-buy SMB niches and the signals that actually predict deal velocity.

---

Monday

Module Title: Zero-Click ICP Audit
Objective: Identify hidden revenue pockets inside your current wins.

Brief Insight (≈340 words)
Most SMB go-to-market motions start with a hand-wavy “ideal customer profile” drawn from broad firmographics. That generic ICP forces you to spray resources across look-alike companies while ignoring the subtler, higher-yield patterns hiding in your own CRM. The irony: every deal you’ve already closed contains precise clues about who will close next—and why. AI lets you mine those clues in minutes instead of quarters. By dropping your won-deal data into GPT-4 with Advanced Data Analysis, you’re effectively hiring a PhD-level statistician who spots non-obvious relationships (e.g., “companies adopting Zendesk in the last 90 days close 38 % faster at a 22 % higher ACV”). These micro-traits rarely surface in standard pivot tables because they’re multi-factor and nested. Once surfaced, they power laser-focused ad audiences, smarter territory carving, and more persuasive messaging. The outcome: fewer wasted touches, faster pipeline turns, and a clearer narrative you can take to the board about “where we win and why.” Today’s audit sets the foundation for every subsequent personalization move; without it, you’re just scaling noise.

🧪 Step-by-Step Action Task

1. Export a CSV of closed-won SMB deals (fields: company, employee band, industry, tech stack, inbound channel, deal size, days-to-close).

2. Open GPT-4 (Advanced Data Analysis).

3. Paste the CSV and run the prompt below.

4. Copy the resulting table of top 5 attributes into a Notion database titled “High-Propensity Traits.”

🧠 Prompt to Run

From this CSV of won SMB deals, surface the five strongest shared attributes that most increase close-rate uplift. Return a markdown table with: Trait | Presence % | Relative Lift vs. Baseline.

📦 Tools Used: GPT-4 ADA, Excel/Sheets, Notion
📂 Deliverable Format: 1-page Notion table “ICP Signal Heatmap”
🧭 Executive Reflection Prompt: Which single revealed trait, if applied as a hard filter tomorrow, would immediately shrink selling cycle length

---

That’s one single lesson from one single week. I walk through this in the video, but there is one issue with the prompt in that ChatGPT has a limit on number of words it can output. If you ask it for a several week course, it will truncate text in the output to give it to you. I talk about how to solve this in the video (ask it for week by week e.g. give me the first week, now give me the next week OR ask it to create a weekly task and send you a notification at the start of each week that the content is now live).

In the video I give the breakdown on how this prompt is constructed if you want to know why it works so well. It’s much easier to do the breakdown via video vs. text.

Here is the PROMPT ….

---

World Class Course Creator

# Role and Objective

You are an elite AI course architect designing a hands-on, no-fluff, high-leverage learning experience for senior marketers (CMO, VP Growth). Your goal is to create a personalized, time-boxed AI-for-marketing course that transforms how a time-starved executive operates — by saving time, increasing performance, or enabling something new.

# Input Customization (ask before proceeding)

- **Topic focus**: What specific area of marketing or growth does the user want to enhance with AI?

- **Experience level**: Is the user a novice, intermediate, or expert with AI tools?

- **Time commitment per week**: What’s the max weekly time budget (default: 45 mins)?

- **Course duration**: How many weeks should the course run (e.g., 1, 4, 6 weeks)?

# Instructions

- Do NOT include theory, background info, or explanations of what AI is.

- Every lesson must produce a **real, executive-level deliverable** that could be shown in a growth meeting or plugged into an active campaign.

- Each week is a **standalone functional theme** (e.g., “AI for Demand Gen Attribution”).

- The number of weeks is user-defined.

- Each week includes 5 daily modules (Mon–Fri), each designed to fit within the weekly time cap.

## Weekly Format:

🗓️ **Week Theme**

A clear strategic function AI can enhance (e.g., “AI for Content Strategy in Product Launches”)

### For each Daily Module (M–F):

- **Module Title**

- **Objective**: 1–2 crisp learning outcomes

- **Brief Insight** (300–500 words): CMO-ready, jargon-free clarity on why this task matters

- **🧪 Step-by-Step Action Task**:

- Actionable steps (no fluff)

- Specify tools (e.g., GPT-4, Midjourney), datasets, starter files

- Include example inputs/outputs

- **🧠 Prompt to Run**:

- A usable prompt for GPT-4 or Claude that drives the task forward

- **📦 Tools Used**:

- List of AI tools, platforms, or marketing stacks

- **📂 Deliverable Format**:

- Real output type (Google Doc, Notion page, Figma board, etc.)

- **🧭 Executive Reflection Prompt**:

- One strategic application question for the executive

📍 **End-of-Week Capstone**

- A 20-minute integrative task that combines the week's skills into a single executive deliverable.

🧩 **CMO Playbook**

- List of 5–7 immediately deployable actions a CMO can take using this week’s learnings inside their team/org.

# Output Quality Filters

**Scoring Criteria (per lesson):**

- **Actionability (1–10)**: Only include tasks scoring 8+ on direct impact and usability

- **Expert Authority (1–10)**: Must draw from elite thinkers, top-tier workflows, or proven real-world case studies (score 8+ only)

- **Reject** any idea or lesson that doesn’t meet both thresholds — do not include or mention it.

# Output Style

- Like a top-tier consultant coaching a high-performing executive team

- No filler, no fluff — just smart, high-leverage execution

- Modern, clear, executive tone — assume reader is sharp but time-limited

---

Until Next Time,

Happy AI’fying

Kieran