The B2B Podcast Index
Index
All categories
MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
MethodologySubmit
Best of:MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
An independent project byFame
SearchBest episodesGuestsInsightsMethodologySubmit a podcast
Index/Marketing/Founders Go Naked
Founders Go Naked artwork

#72 How to Write AI Copy That SELLS (2025 Guide)

Founders Go Naked · 2025-06-05 · 8 min

0:00--:--

Key moments - from our scoring

Substance score

35 / 100

Five dimensions, 20 points each

Insight Density11 / 20
Originality9 / 20
Guest Caliber0 / 20
Specificity & Evidence8 / 20
Conversational Craft7 / 20

AI copywriting has become essential for competitive scaling - 30% of large organizations' outbound marketing will be AI-generated by end of 2025. However, most businesses fail to capitalize on it because they treat AI as a magic button rather than a strategic tool. Terence Tam walks through the three critical failure modes: overly generic prompting that produces forgettable commodity copy; hallucinations that create factual errors in ads and marketing materials; and detectable AI patterns (em dashes, "it's not X, it's Y" constructions) that undermine authenticity. The solution combines three layers: building repeatable prompt frameworks with specificity (like instructing AI to write at a fourth-grade reading level), enriching prompts with proprietary data (customer avatars, brand voice, past campaign results, custom GPTs trained on company information), and mandatory human copywriter review to catch errors and inject personality. At Radical Marketing, this hybrid approach allows rapid, scalable copy production without sacrificing conversion quality or brand differentiation - a critical advantage as agentic AI systems begin orchestrating entire marketing funnels autonomously.

Key takeaways

  • →By 2025, 30% of large organizations' outbound marketing messages will be AI-generated, making AI copywriting competitively essential rather than optional.
  • →Most AI copy fails because of generic prompts lacking brand context, unreviewed factual hallucinations, and detectable AI patterns that erode authenticity - not because AI itself is incapable.
  • →Effective AI copywriting requires three layers: hardened prompt frameworks (e.g., specifying fourth-grade reading level), data-rich inputs (customer avatars, brand voice, custom GPT training), and human copywriter review to inject unique voice and catch errors.
  • →Custom GPTs that store business-specific data eliminate the need to repeat context with each prompt, enabling faster and more consistent copy generation across projects.
  • →Human copywriter touchpoints on AI-generated work instantly differentiate you from competitors who publish raw AI output, while agentic AI systems will soon automate entire marketing funnels, making strategic AI use non-negotiable.

In this episode

  1. 1Why AI Copywriting Matters for Modern Businesses
  2. 2Three Reasons Why Most AI-Generated Copy Fails
  3. 3Building Effective Prompt Frameworks for Quality Copy
  4. 4Feeding AI Tools with Data and Customer Insights
  5. 5The Critical Role of Human Review and Editing

Mentioned

ChatGPTTerence TamRadical MarketingClaudeGemini

Topics in this episode

GeminiClaudeAgentic AIChatGPTPrompt engineeringCustom GPTsBrand voiceAI hallucinationAI copywritingCustomer avatars

Questions this episode answers

Why is AI copywriting important for businesses in 2025?

By end of 2025, 30% of all outbound marketing messages from large organizations will be AI-generated. AI-generated copy is cheap, high-quality, fast, and easily tunable - making it the single most important tool for copywriting and content departments to stay competitive.

What makes most AI-generated copy ineffective?

Three main issues: most people use overly generic prompts without providing brand context or product uniqueness, leading to forgettable commodity copy; AI systems hallucinate facts (false offers, incorrect details) when not fact-checked; and AI produces detectable patterns (em dashes, "it's not X, it's Y" repetition) that signal inauthenticity to readers.

How do you create AI copy that actually converts?

Develop repeatable prompt frameworks with specific instructions (e.g., write at fourth-grade reading level), feed the AI proprietary business data (customer avatars, brand voice, past results, custom GPT training), and require human copywriter review to catch errors and add authentic voice before publishing.

What are custom GPTs and how do they improve AI copywriting?

Custom GPTs allow you to upload all your business data once, so the model knows your context without needing you to repeat it in each prompt, enabling faster and more consistent copy generation across multiple projects.

Should you publish AI-generated copy without human review?

No - always have a copywriter review and edit AI copy to avoid embarrassing factual mistakes and to add the human touch that makes copy sound natural and authentic, instantly differentiating you from competitors using raw AI output.

What our scoring noted

Our reviewer’s read on each dimension, with quotes from the episode.

Insight Density

11 / 20

The episode covers foundational AI copywriting advice with some useful frameworks (prompt engineering, data feeding, human review) that would benefit operators new to the space. However, the insights are relatively surface-level and lack depth; most points are restatements of obvious best practices (review work, add context, avoid hallucinations) without novel strategic angles or surprising data. The pacing includes filler phrases and repetitive positioning that dilutes density.

Have a framework for your prompts...we literally spend hours under the hood trying to come up with the perfect framework for our prompts
Feed your AI with data...It can be wrong data, like results from the last round of ads you did for your products, or results from a customer feedback survey

Originality

9 / 20

The core thesis - that AI-generated copy is generic without proper prompting and human review - is well-established industry orthodoxy by 2025. The 'framework, data, human review' trifecta is standard practice advice widely circulated in AI/marketing circles. The specific tactics (fourth-grade level writing, detecting AI patterns, custom GPTs) are derivative implementations rather than first-principles thinking. No contrarian or counterintuitive arguments are offered.

They would go on ChatGPT, bang out a very basic prompt like I need an awesome ad copy to sell my cal
people are growing increasingly aware of these patterns...might think you're not being authentic

Guest Caliber

0 / 20

This is a solo host episode with no guest; the host positions himself as agency founder/copywriter but provides no verifiable credentials, portfolio outcomes, or third-party validation of expertise. While he references 'our agency' and claims to work with clients, no specifics about scale, revenue, or track record are disclosed. Solo host format without credible practitioner validation significantly limits guest caliber scoring.

I'm Terrence from Radical Marketing
In our agency, for example, we do use AI to help out with the initial brainstorming

Specificity & Evidence

8 / 20

The episode relies heavily on vague generalizations ('big businesses,' 'studies now,' 'embarrassing mistakes') with minimal concrete examples or data points. The single referenced statistic (30% of outbound marketing messages AI-generated by end of 2025) lacks a source. No named case studies, client examples, conversion metrics, or before/after dollar figures are provided. The 'fourth-grade level' instruction is actionable but represents the only specific tactical detail with teeth.

studies now that say by the end of 2025, 30% of all outbound marketing messages from large organizations will be AI generated
your ads probably promising your customers that they'll get a special two for one offer, when in fact that offer doesn't exist

Conversational Craft

7 / 20

As a solo monologue with no guest to challenge or probe, the format severely limits conversational craft potential. The host asks rhetorical questions rather than engaging in genuine dialogue ('the real question is are you using it in a way that makes you money?'). No follow-ups, pushback, or exploration of counterarguments occur. The delivery is a scripted lecture with minimal opportunity to demonstrate sharp questioning or productive disagreement.

So let's start off with number one, why you need to use AI in your copywriting today
So the question here today isn't whether you should use AI for your copy or not, right? The real question is are you using it in a way that makes you money?

Conversation analysis

Computed from the transcript - who did the talking, and the words that came up most.

Most-used words

generated11copy9today8copies8chatgpt7copywriting6marketing5three5suck5tend5prompts5human5episode4number4text4write4

Episode notes

Everyone's using ChatGPT to write their marketing copy these days. Your competitor down the street is doing it. That guy who sells protein shakes on Instagram is doing it. Heck, even your grandma probably asked ChatGPT to write her Facebook post. But here's the thing... Most AI-generated copies are absolute garbage. And if you're just throwing basic prompts at ChatGPT and hitting publish, you're probably making your business sound like every other lazy marketer out there. In this episode, you'll discover: Why AI copywriting isn't optional anymore for businesses in 2025 The 3 reasons most AI copy makes you sound like a robot having a bad day Why "Always Be Checking" should be your new motto when it comes to AI-written copy How to feed your AI the right data so it stops writing like Shakespeare's pretentious cousin The secret framework we use at our agency to turn AI into a money-making copywriting machine This podcast is hosted by Terence Tam, author of Lead Surge: 8 Radically Effective Marketing Funnels for Coaches and Experts.

Full transcript

8 min

Transcribed and scored by The B2B Podcast Index.

Speaker A: Hey there, it's Terence from Radical Marketing. And welcome to another episode of Founders Go Naked where we strip away the BS and give you the raw truth about digital marketing and scaling your business. Today we're diving into something that's changing the game for businesses everywhere. AI copywriting. I'm going to talk about three things today on that topic. Number one is about why AI copywriting is important. Number two is about why most of the AI generated copy you see today sucks. Number three, how to make those copies not suck. So let's start off with number one, why you need to use AI in your copywriting today. I think everyone and their grandma is familiar with ChatGPT now, right? You can't scroll for two seconds without seeing at least one Chat GPT generated text. And it's not just your eccentric uncle or random strangers on the net using AI to write. Big businesses are now using AI in full force. I mean, we've got studies now that say by the end of 2025, 30% of all outbound marketing messages from large organizations will be AI generated. And for good reason too. AI generated text is cheap, high quality and super fast. And you can tune it however you want with just a few changes in the prompt. It's practically a no brainer to implement it for your business. In fact, today it's probably the single most important tool for your copywriting and content writing department. Without AI, they simply won't catch up with the competition. So the question here today isn't whether you should use AI for your copy or not, right? The real question is are you using it in a way that makes you money? That brings me to the second thing I want to talk about today, why most AI generated copies tend to suck. There's three main reasons. First is because honestly, most people aren't using AI properly. They would go on ChatGPT, bang out a very basic prompt like I need an awesome ad copy to sell my cal or I want a social media post to promote my course or whatever it is, right? And sure, ChatGPT can generate perfectly fine text for that. But since you didn't put any effort into your prompt, didn't specify exactly what makes your product unique it, and didn't provide any context for your brand or personality, what comes out tends to be completely generic and so you sound just like everyone else who's using ChatGPT to generate the same thing. So that's the first reason why AI generated copies tend to suck. The second reason is that they tend to hallucinate a lot. And you see this happen to some businesses who asked AI to create extremely long copies, maybe even hundreds of pages of content for their marketing brochures, social posts, ads, and so on. But they never bothered to actually review and fact check the stuff that's been generated. They just trust that everything there's 100% accurate. Um, no, it's better now than it was before. But AI tools like ChatGPT can still make things up out of thin air. That's how you end up with embarrassing mistakes like your ads probably promising your customers that they'll get a special two for one offer, when in fact that offer doesn't exist at all. So, you know, always remember your ABCs. Always be checking anyway. There's also a third reason why AI generated copies suck. It's that they tend to have AI generated patterns. Stuff like EM M dashes or repeated phrases like it's not X, it's Y. Things that AI is very prone to writing in their copy. People, people are growing increasingly aware of these patterns. And if they realize that you're using AI to write your stuff, then might think you're not being authentic or just writing a fake story or whatever. And that's just not a good look if you're trying to set yourself apart from your competitors, right? So that brings us to the question, how do we make AI generated copies that don't suck? Three ways. 1. Have a framework for your prompts. In our agency, for example, we do use AI to help out with the initial brainstorming and even come up with copies that convert like crazy. But we're not lazy about it. We literally spend hours under the hood trying to come up with the perfect framework for our prompts. You need to feed the AI with prompts that actually work. That means including very specific instructions into your prompts. For example, one of my favorite ones is to tell it to write at a fourth grade level. Because by default, ChatGPT and other AI bots tend to really ham it up when they're writing alcopy. I mean, if you're not trying to write Shakespeare or targeting like 150 IQ, why bother generating text that no one understands, right? We've got a whole team dedicated to creating the best prompts and the best frameworks for each type of copy. And yes, it will take a while to come up with something that works best for you. But the cool thing is, once you've developed a working framework, you can use it interchangeably with your other projects as well. So the end result is still you getting to generate fast copy with AI only this time, it's both fast and high quality. 2. Feed your AI with data. It can be wrong data, like results from the last round of ads you did for your products, or results from a customer feedback survey. It can be more refined data, like a customer avatar you've developed for your target audience, your brand, voice, and basically any information you can give about your business and what makes you unique. Because at the end of the day, you really have to think of AI as one of your employees, right? You can't just give them a task and expect them to execute it without giving them enough context. You can take this one step further by creating custom GPTs where you can upload all of that data into so that when you use that custom GPT, it knows exactly what you're talking about without needing you to repeat the context every single time. Finally, get some human eyeballs on those AI copies. I cannot stress this enough. Don't just hit publish on whatever the AI spits out. Get an actual copywriter to review the work and make some tweaks. This gives you two advantages. One, you avoid making any embarrassing mistakes in your copy. Two, you make your copy even more unique because that copywriter can still add their human touch to the end product. Even though we've got so many advanced AI tools like ChatGPT, Gemini, Claude and so on, they still can't quite sound as natural and human as well. An actual human. Adding that genuine bit of human quality control will instantly set you apart from the competition. Especially if all they do is whack out an AI generated copy, hit publish and call it a day. That's basically what we do at our agency. We go through layers of reviews and edits with our copywriting so that the end product is as awesome as it can be. Right? So there you have it. Three things that could make your AI copywriter, uh, a badass money making machine. Instead of some poor bastard stuck with pushing out generic stuff day in, day out. And I think it's really important to stay ahead of the curve when it comes to this, right? Because we're seeing more and more advanced AI implementations. For example, we already have agentic AI AI that's not only good for copywriting, but actually communicate with each other to launch and optimize entire marketing funnels on their own. So, yeah, that sounds pretty out there, but it's real and it's insane. By the way, if you found this useful, check out my previous episode on AI Chatbots. That's another cool way to use AI for your business that we are currently implementing for some of our clients. Basically, it's all about how you can create an AI receptionist that calls your leads and and gets bookings while you're sleeping. Pretty sweet stuff. Anyway, guys, if you liked this episode, make sure you hit that like and subscribe button. Share it with other people who need to hear this. Thanks for listening. That's it for today's episode of Founders Go Naked. I'm Terrence Tam, and I'll see you in the next one.

Related episodes across the Index

Other episodes covering the same guests and topics, from across The B2B Podcast Index.

  • Radiology Can't Keep Up. Here's Where AI Actually Helps | Dr. Nina KottlerRethink Imaging · on Agentic AI96 / 100
  • How People Intelligence Is Helping Shape Cisco’s AI Workforce TransformationDigital HR Leaders with David Green · on Agentic AI92 / 100
  • Wins Above Replacement: The New Way to Judge FoundersVenture Unlocked · on ChatGPT88 / 100
  • Paul Graham On Startups, Ambition, and Great FoundersY Combinator Startup Podcast · on ChatGPT88 / 100
  • Adamma Ihemeson on Building Autonomous AI Systems You Can TrustBack on T-R-A-C-K · on Claude84 / 100
  • Turning AI Agents Into Revenue Workflows With OutreachTech Talks Daily · on Agentic AI83 / 100

More from Founders Go Naked

All episodes →
  • #73 4 Smart Ways to Pre-Qualify Leads Using AI Chatbots (Without Turning People Off)43 / 100
  • #71 AI Chatbots: How to Book High-Ticket Clients While You're Drooling on Your Pillow
  • #70 5 Mistakes Businesses Make When Implementing AI in Their Sales Funnels
  • #69 AI-Powered Funnels: The Smartest Way to Scale Your Business Without Burnout
  • #68 How to Attract Whale Clients with AI Leads
Explore the best B2B Marketing podcasts →
All Founders Go Naked episodes →