25 April 2024

I Created An Ai Email Course Using ChatGPT

2Frameworks

8Insights

Frameworks in this episode

Marketing7 steps

Customer Language Grammar

Turn customer conversations into an AI-powered copy quality system

Read →

Productivity6 steps

Five-Pass AI Iteration Loop

Make AI critique and improve its output five times before human editing

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster14:30

AI Avatar Videos Educate Better Than They Persuade

The course used HeyGen to turn AI-generated scripts into videos featuring Flanagan's avatar. He found the format efficient and acceptable for short educational summaries, but too formulaic and emotionally flat for videos intended to sell, convert, or create a strong feeling.

Keeping physical gestures limited improved avatar realism
AI avatars remove the need to record every lesson manually
Short educational videos are a practical use case
Limited emotional range weakens persuasive video

“I did really nothing other than like again continue to prompt AI where I thought it could do better”
Kieran Flanagan · 14:30

“if you really want the video to be something that sells people on something or converts them or makes them feel a certain way then…”
Kieran Flanagan · 16:30

#ai video#heygen#education#conversion

Hot Take· 1

Hot Take24:30

Identical Prompts Degraded in ChatGPT While Claude Finished the Job

A previously functioning homework prompt began returning instructions for the user instead of executing the requested campaign workflow. Flanagan described ChatGPT as increasingly lazy in this test, while Claude followed the same core request and completed five progressively developed campaign iterations.

A prompt that previously worked changed behavior over time
ChatGPT delegated requested work back to the user
The model identified potential improvements but did not apply them
Claude completed the five requested iterations

“chat gbt is getting really lazy and is getting answers are getting much worse than when they were like two to three weeks ago”
Kieran Flanagan · 24:30

“Claude does a much better job than chat gbt right now”
Kieran Flanagan · 34:30

#chatgpt#claude#model reliability#prompt degradation

Explainer· 2

Explainer02:30

The Three Types of Marketer Responding to AI

A highly discussed LinkedIn post revealed three broad reactions to AI among marketers: outright dismissal, narrow use as a writing assistant, and recognition that it will fundamentally change work. Flanagan positions the podcast as a resource for the smaller group willing to experiment early and become future leaders.

Some marketers dismiss AI as hype
Many restrict AI to assisting with blog posts
A smaller group expects AI to reshape marketing work
Early adoption can create an opportunity to move ahead of peers

“marketers really fell into three categories”
Kieran Flanagan · 03:00

“we want those marketers to be the Risk Takers the opportunistic the people who do things first the present and future leaders”
Kieran Flanagan · 03:30

#ai adoption#marketers#innovation#career strategy

Explainer31:00

The Next AI Course Could Combine Experts, Animation, and Community

For a more advanced version, Flanagan would add an interactive AI representation of himself, licensed expert avatars, richer animation, and a community bot. He believes this combination could make courses more engaging while allowing creators to remain useful to a community without personally answering every interaction.

An AI version of the creator could remain available throughout a course
Expert avatars could teach lessons tied to their recognized expertise
Animated videos could reduce the monotony of repeated talking-head lessons
Community bots could increase interaction without consuming all of the creator's time

“I would create more animated videos”
Kieran Flanagan · 32:00

“I think Community Le with courses might be a big thing because I think you can create them with a bot AI version of yourself”
Kieran Flanagan · 32:30

#online courses#ai avatars#community#animation#education

Story· 1

Story07:00

An AI-Written Landing Page Reached a 43% Conversion Rate

Flanagan used AI to draft and revise the landing page for his six-day marketing course, including copy influenced by classic copywriter John Caples. With roughly 500 signups at the time of recording, he reported a conversion rate of about 43%, while stressing that he mostly directed and lightly edited the AI's work.

AI generated the landing-page copy
The process explored multiple variations
The offer combined classic marketing lessons with modern AI prompts
The reported conversion rate was about 43%

“the landing page was created using AI right”
Kieran Flanagan · 07:00

“the conversion rate to date on about 500 signups is about 43%”
Kieran Flanagan · 08:00

#landing pages#conversion rate#ai copywriting#email course

Tool· 1

Tool25:30

Use PDFs to Route Complex Custom GPT Instructions

Flanagan encountered an 8,000-character instruction limit while building a multi-lesson homework GPT. His workaround was to place detailed lesson instructions in uploaded PDFs and keep the main instruction field as an if-then router that selects the correct PDF based on the user's choice.

Complex custom GPTs can exceed the main instruction limit
Detailed workflows can be stored in uploaded PDFs
The primary prompt can act as an if-then routing layer
Each user selection triggers the corresponding instruction document

“the custom GPT has an 8,000 character limit so you can only actually incorporate 8,000 words of instructions”
Kieran Flanagan · 25:30

“what I put into the custom gbt itself in terms of the instructions is just a big if then else Loop”
Kieran Flanagan · 25:30

#custom gpt#prompt engineering#pdf#course tools

Takeaway· 2

Takeaway01:00

Why Small Side Projects Keep Marketers Close to the Craft

Flanagan argues that small, hands-on projects are one of the best ways to learn regardless of seniority. Shipping the email course also let him experience the direct satisfaction of creating something useful instead of only managing people, boxes, and budgets.

Side projects create practical learning opportunities
Senior marketers can become detached from hands-on craft
Real signups provide immediate feedback on whether work is valuable
The course attracted roughly 450 to 500 early signups after limited promotion

“you should always have little side things that you do it is the best way to learn”
Kieran Flanagan · 01:00

“it's really how I learn I have to actually do the thing”
Kieran Flanagan · 02:00

#learning#side projects#marketing craft#career growth

Takeaway29:30

Flanagan's Scorecard for AI Course Production

Flanagan rated AI as competent but not exceptional across most stages of course production. He gave the outline, format, and emails roughly three out of five, while advanced prompts created with Claude earned four out of five; his conclusion was that AI can produce a valuable course for many small businesses, though probably not the internet's best course on a subject.

AI handled the outline, format, emails, scripts, and landing page
Most course components earned roughly three out of five
Claude's template-guided advanced prompts earned four out of five
The result was useful but not positioned as category-leading

“showline I would give it like a three out of five did really well episode format three out of five”
Kieran Flanagan · 30:00

“would it be the best course on that topic on the entire internet no but would it be pretty good that people will find valuable…”
Kieran Flanagan · 30:30

#course creation#ai quality#scorecard#small business