MMarketing Against The Grain
← All episodes
05 December 2024

Copy ANY Writing Style with AI in 30 Seconds (Step-by-Step)

2Frameworks
7Insights

Frameworks in this episode

Insights & moments

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

Myth Buster· 1

Myth Buster26:30

AI Multiplies Expert Output Rather Than Replacing Expertise

The episode rejects the idea that AI instantly turns inexperienced people into excellent writers. Its larger advantage goes to people who already possess domain expertise and editorial judgment, because they can recognize quality, direct the model, and produce much more work.

  • Domain experts can provide stronger inputs and judge outputs more accurately.
  • AI does not automatically create high-level writing skill.
  • Experienced writers can use AI to increase production dramatically.
  • The main near-term advantage is multiplication of expertise.

AI is a boom for people with real domain expertise.

Kieran Flanagan · 26:30

It's for people who are good at writing to do 10x more.

Kieran Flanagan · 26:30
#domain expertise#productivity#writing#ai leverage

Hot Take· 2

Hot Take22:00

Could Personalized Niche Fiction Make £10K–£20K a Month?

The hosts propose researching underserved fiction genres through fan forums, producing pseudonymous books with AI assistance, and paying for professional editing. A companion web app could personalize purchased books, capture reader emails, and reduce dependence on Amazon through direct remarketing.

  • Find genres with strong demand and relatively little competition.
  • Research audience preferences in niche forums and fan communities.
  • Use AI for drafting while retaining professional editing.
  • Offer lightweight personalization as a digital add-on.
  • Capture emails to build a direct customer relationship.

I bet you could make 10-20 grand a month just doing that.

23:00

You can build, you could build a really sick little funnel of leads

24:00
#fiction business#personalization#amazon#ai publishing#funnels
Hot Take26:30

Small AI Writing Businesses Could Run in One Evening a Week

The hosts speculate that lightweight AI-assisted businesses could be operated in focused two- or three-hour weekly sessions. Examples include fiction publishing and simple gated web applications, with AI providing enough productivity leverage for one person to test several small revenue streams.

  • Use AI to compress production into short focused sessions.
  • Choose lightweight products such as fiction or gated web apps.
  • Operate different experiments on different nights of the week.
  • Reserve human effort for expertise, editing, positioning, and judgment.
  • Treat the revenue estimate as an entrepreneurial hypothesis, not a guarantee.

I literally think we could pick a night of the week and have a different business that we run just that night of that week

Kieran Flanagan · 26:30

you could make 20 to 50 grand a month with just some basic work.

Kieran Flanagan · 27:00
#side business#solopreneurship#ai productivity#micro-saas

Explainer· 1

Explainer11:30

ChatGPT Extracts Nuanced Style Guides but Struggles to Apply Them

ChatGPT produced a more detailed writing guide than Claude's automatic style feature, identifying tone, humor, sarcasm, structure, and paragraph functions. However, its first attempt to apply that guide to a HubSpot company breakdown remained surface-level, showing that analysis quality does not guarantee generation quality.

  • ChatGPT identified subtle qualities such as respectful humor and light sarcasm.
  • Its guide included structural functions for different parts of an article.
  • The generated company breakdown did not match the quality of the extracted guide.
  • A strong style guide may work better when transferred into another model.

It did create a better, more detailed style guide than Claude.

Kieran Flanagan · 12:00

pretty great for creating a writing style guide that you could then go use in Claude.

Kieran Flanagan · 16:00
#chatgpt#claude#model comparison#creative writing

Story· 1

Story20:00

A Podcaster Screenplay Test Exposes AI Fiction's Limits

One host asked ChatGPT to write a screenplay about two long-distance podcasters discussing AI and marketing. The model reproduced surface details and familiar show dynamics, but the hosts judged the screenplay itself as poor, illustrating the gap between recognizable imitation and compelling fiction.

  • The prompt included locations, friendship dynamics, and recurring audience criticism.
  • ChatGPT generated a dramedy titled Podception.
  • The screenplay echoed the show's cold-open style.
  • Accurate surface details did not make the script creatively strong.

It wrote Podception, a dramedy and it's great.

20:30

It wrote a pretty awful screenplay, which I think is not really good at all

21:00
#screenwriting#fiction#chatgpt#creative writing

Tool· 2

Tool01:00

Claude Offers Fast Style Creation and Deeper Manual Control

Claude can create a custom style from a pasted writing example or from a manual description. The automated route is fast and usable, while manual editing allows more detailed principles, formatting rules, hooks, and closing patterns.

  • Create a style from a writing example in about 30 seconds.
  • Use manual descriptions when greater control matters.
  • Detailed guides can specify sentence length, paragraph structure, hooks, and endings.
  • Treat the generated style as a starting point rather than a finished system.

you can basically add a 'writing example', or you can 'describe this style'

Kieran Flanagan · 01:30

I think this is how you get a better just better results: Edit style manually.

Kieran Flanagan · 05:30
#claude#style guides#ai writing#prompting
Tool24:00

AI Makes Free-Form Customer Answers Operational Again

Free-form answers once created a difficult analysis problem because teams could not group them reliably at scale. The hosts argue that AI can now extract, cluster, and use those answers for personalized nurturing, sales follow-up, and re-engagement of closed-lost opportunities.

  • Ask customers or prospects about their biggest challenge in their own words.
  • Extract relevant details from discovery calls automatically.
  • Cluster semantically similar problems without restrictive dropdowns.
  • Use the grouped insights to personalize nurturing and follow-up.
  • Apply the same data to closed-lost re-engagement.

just that free form text alone allowed us to, on the backend, personalize things at scale.

24:00

you don't need to use drop down. It can just be a free-form text because AI can actually group all that stuff together.

25:30
#customer data#segmentation#sales#nurturing#ai automation