While I am wholly against using ChatGPT for reasons I don’t want to waste anyone’s time over-explaining (here are some resources though on how unethical the company is, the environmental impact, and how our government is using ChatGPT in terrifying ways), I do feel more open to Anthropic both for its ethics based approach and its advanced capabilities that are helpful to both myself and my clients.
This article is NOT about the ethics of AI, because I am not an AI-ethicist, and because the technology is evolving so rapidly that I feel my opinion on the subject changes as I see how these companies grow and change.
That being said, personally, I see the difference this can make in business and have been using it as such. However, it is often reductive, misleading, and frankly basic. I have actual case studies at this point in my business and the businesses of my clients about where it may save you time, but is actually costing you money.
And that arena, specifically, is in copywriting. (At the bottom of this post I give you the actual prompts I DO feel AI is incredibly helpful in using.)
Claude is a great brainstorming tool. A helpful editor. An aggregator of data. But, it can’t make you content that doesn’t sound like everyone else’s content, because AI pulls from what exists. (Or at least Claude does, ChatGPT I feel like occasionally just makes up garbage…though that is my anecdotal opinion).
Perhaps you’ve noticed it, without consciously realizing it, but all of a sudden all the hooks and the subject lines have started to sound the same:
The curiosity gap: “The thing no one is talking about…that you didn’t even know was holding you back.”
The shock factor: “I almost didn’t tell you this…”
The social proof: “I made $X in five min using AI”
False depth priming: “You think it is about X, but it is about Y”
Single relatable, signal confident, signal a quick promise of results.
This is not copywriting—it is mad-libbing. (I wrote that m dash myself, thankyouverymuch.)
And then when everyone sounds like everyone, (personally I find it boring) it is not good for marketing (especially in the service provider space) because it subconsciously makes your customer sus. Cuz good marketing is about human-to-human connection. It makes you feel something. Marketing that sounds like everyone else’s marketing reminds your customer who is looking for something that actually could work for them of the same sales page as the thing that they wasted their money on…
And a robot can’t concoct a human-to-human connection.
As someone who spends a good chunk of my week line-editing my clients’ front facing marketing, the repetition from AI use has started to really irk me. (If you are a client this is not personal!!)
What it scratches at is that despite working with different business owners, in different phases of growth, with different sized lists, and different ideal clients…their marketing sounds…the same.
The more prevalent AI has become, the more dramatically I see the replicable formatting get more and more obvious. Can you relate to this?
Now at the risk of sounding like a raging hypocrite…here is some real tea for y’all…
In May of 2026 I had Claude write all of my sales emails.
From March -May of 2026, my grandmother died and then my dog died. I was sleeping on the floor in a giant dog bed getting a fragmented 3 or so hours per night as Addy’s health rapidly deteriorated. And I launch my main course in January, May and September.
These launches bring in over half of my revenue. I also really enjoy the community and did not want to opt out of doing it. Additionally, I have launched 10 times now, so the majority of the process is sort of rinse and repeat. I know with a very small margin of error at this point how many people will sign up.
The main thing that is fresh every time? The copy I write in the emails. And being that sleep deprived, I did not have it in me.
Instead I took all of our best performing emails from the last handful of launches, fed them to Claude, and asked it to riff off of the subject matter, mimic my tone, and construct the launch emails I needed according to my content calendar.
Then I read through and edited them to sound more like myself. I knew they were’t as strong. I knew they didn’t nail the emotional cohesion and depth I do. And I knew I did not have it in me to generate creativity. I was completely emptied. (So maybe, just maybe if you are in a hard season of life you can use it for this, but let’s also hold I have an extremely well cultivated warm audience and I taught a bomb ass webinar, so this might not work for someone else.)
The results?
We converted 3.8% of the people who signed up for the webinar into purchases of the program. (The national average for webinar conversion by the way is ~2%)
Usually, I convert at 8-12%. So with AI doing the writing, that number, though still comparatively high to the national average, was more than halved.
Sure, we could hypothesize that my state of mind at this time also contributed to this result, but email marketing is our number 1 driver of purchasers with almost a decade of data to prove that. And this was still an over $70k launch so I, considering what was going on in my life, consider this a success.
That being said, I will die on the hill that providers not only shouldn’t but can’t really write their most effective marketing content with AI.
Most recently, I fell for a trope-y Substack article about 3xing your IG following using Claude as a content strategist and calendar creator. I thought, “This isn’t copy, this is brainstorming! Let me try it!” Especially as constantly producing content is something that I find draining.
I followed the prompts laid out in the article and hated every single idea the platform fed me.
First, I asked it to do market analysis of my niche…and it pulled up other anticapitalist business coaches—all of whom had smaller audiences than I and were advertising specifically that they make was was not even half of my revenue.
I said to Claude, “Why would you use businesses not as successful as mine as an example of what we should try? These people are my future clients, not my example.”
“Good flag,” it responded, “Let me pull more successful examples.”
Then it spit out stuff that sounded like the very girl-boss-gobbily-gook I actively try to position myself in opposition to (despite having been given my entire Substack, website and content history as an example before being given the assignment.
“I am not a pseudo-liberal ‘make $10k a month in your sleep’ girlie…where did this even come from?”
“Good flag—that is actively not your brand voice. Let me do another go around.”
Then it tried to position me among the likes of Rachel Cargle, Adrienne Maree Brown, and The Nap Ministry. To which I actually lost my shit and said in all caps, “WHY WOULD YOU CO-OPT THE IP OF BLACK ACTIVISTS AS IF I SHOULD USE THAT AS A WAY TO POSITION MY ANTICAPITALIST BUSINESS CONTENT. WHY ISN’T THIS WORKING?”
Here was it’s response:
On the real problem — “everything is either politically inappropriate, based on people less successful than me, or waters down my message” — here’s the actual pattern from this whole session: every time I brought in an outside creator as a reference or model, it failed. Too small, appropriative, too girlboss, wrong politics, whatever the specific reason. Meanwhile every time something actually landed — it came from your own real material: your own site copy, your own account’s engagement data.
So here’s what I’d actually propose: stop looking for a comparative creators entirely. Build the whole calendar from only two sources — your own past top-performing content (real data, already pulled) and your own real, unpublished material (stories, site copy, actual language you use) that you verify as accurate. For technique, use only unattributed structural patterns — hook mechanics, CTA structure, pacing — as abstract craft knowledge, not tied how any one else is doing it.
What I took from that? My human brain creates better and more resonant ideas that are reflective of my work because I am one-of-one and I cannot use an existing framework to grow.
And frankly, neither can you.
I was just talking about this in my group coaching program Feminist Business Framework this week.
The problem with most coaching programs is that they sell you some kind of replicable template—often what they did as if it should work for you. And if/when it doesn’t, they blame your mindset in lieu of looking at how their strategy might not be universally applied to just-about-anyone.
I think this is why so many people leave containers they feel they have wasted a ton of money on feeling disheartened and gaslit instead of with the results promised on the sales page.
(Just a note that I also think the public perception of building an online coaching business makes it seem like it supposed to be easy, and many people are gonna shoot their shot when it ultimately isn’t for them and those people also leave coaching containers blaming the facilitators in lieu of taking accountability that their business strategy either did not have a viable audience or structure and would have never been successful…ijs…)
That being said…I still think that Claude can do AMAZING things for your business like:
Identify ideological inconsistencies in your branding (ie. you call the offer one thing at the top of the page and something else at the bottom and you didn’t even notice)
Point out where you may lose traction because of lack of clarity in working, grammar and syntax
Brainstorm ideas with you (even tho most of the ideas, imo are terrible) so that you get kicked in the right direction of coming up with something on your own
Taking your data and giving you good inferences about how to interpret it
Building you launch calendars and to do lists
Pulling research of your ideal clients across social platforms and giving you a comprehensive analysis of how your customer describes their core issues in their own words (invaluable)
If you want to get the actual prompts I do use to get Claude to provide assistance in business in ways that actually work…
