How to Become a Dangerously Good Writer (with AI)

Source: ActiveCampaign•

How to Become a Dangerously Good Writer (with AI)

Copywriter Justin Oberman was scared AI would replace him. Instead, it’s made him more creative (and busier) than ever.

  • Blog
  • Copywriter Guide Ai

October 30, 2026

When ChatGPT went mainstream, Justin Oberman, like many copywriters, went through an identity crisis.

After being spooked by the prospect of his trade disappearing, he tried out different ways of positioning himself, including as a publicist, a personal brand strategist, and an “impresario.” Ultimately, he decided to double down on copywriting.

And it’s paid off.

Justin has been a copywriter for more than 20 years. He cut his teeth at advertising agencies before starting his own in 2020. He gained 10,000 LinkedIn followers by posting daily, and as his client base grew, so did his revenue: seven figures within three years.

In 2024, he shut his creative agency down to pivot to executive ghostwriting for high-net-worth business owners—and he’s never been busier.

Justin still loves to do a lot of the work on paper. He’s no AI purist, either: he uses it for drafts. What sets him apart is where he puts his own effort: the thinking and the editing, which is where great writing happens.

His rule: “The last step should be human, not AI.”

Here’s how a copywriting veteran writes human copy in a post-AI marketing world.

The approaches:

Teach AI your frameworks and mental models

Eugene Schwartz was a direct response copywriter who wrote one of the most significant books on copywriting, Breakthrough Advertising. One of its core ideas is mass desire, which is the “public spread of a private want.” In other words, the hopes, dreams, fears, and desires that already exist in the hearts of lots of people (like wanting to look and feel good or having enough money to retire).

Before writing a word of copy, Justin needs to understand the “mass desire” around his client’s product. After all, if you're talking to the wrong people, it won’t be interesting, and it won’t matter what you’re saying. AI can do the first pass on who your audience is and what they want to hear.

Justin combined notes from Breakthrough Advertising and instructions that give him a starting point for every piece of copy

Justin summarized the first four chapters of the book and put his notes into a Claude project that does deep research for his ads. It comes back with where his audience sits on two scales: awareness and sophistication.

Eugene Schwartz’s mass desire framework

How well people know your product, category, or brand

How many similar claims the market has already heard

Then it looks for existing ads and social media conversations and gives him a report. Then he checks it against the real world, because AI "really is just looking for the average." If he were selling pencils, that would mean going to Staples and watching people buy them.

Justin’s approach combines conversations with AI, real-world research, and a knowledge base that organizes all of it. Over the years, he’s built the project into a multistep process that runs for the better part of a day.

Inside the project:

  • awareness stages
  • market sophistication stages
  • desire stages
  • lead types
  • hooks that belong to each one
  • offer proof mechanisms

When it's time to do research, he uses this prompt:

Using the Breakthrough Advertising notes in this project, assess the market for [product] sold to [audience]. Tell me (1) the dominant awareness stage and (2) the sophistication stage, with the evidence for each. Then tell me what that combination means for the headline. Flag anything you're guessing.

Justin’s Breakthrough Advertising project instructions

Before AI, this kind of research could take days of reading to put together. Now the desk research is done in hours. In the end, he has a deep understanding of who he’s talking to and what he needs to say.

For you: What frameworks do you follow in your craft? Giving the AI the systems you use can help your work stay consistent and automate some of the more repeatable parts.

Do research on outliers to get outsized engagement

Once Justin has the mass desire clinched, he uses Eden to research the top creators and find what topics and messaging the market is responding to. The platform lets him filter by follower and

Justin looks at Substack and YouTube titles and headlines from creators because they’re easiest to absorb. Or he’ll search within topics like business or self-improvement.

“I'm basically looking at the mass desire of people, of the world,” he says. “Here's what people are really into within this category or in general.”

At this stage, search for both the product (say, “pencil”) and the desire. Look for outlier content from top creators: YouTube videos, ads, Substack posts, and LinkedIn posts that perform at least 10 times better than average.

For you: AI is an always-on research assistant. You can have it gather successful campaigns, designs, sites, and other “swipe files” for you to learn from.

Put your own spin on the outliers

Once Justin’s found outliers around his topic, he thinks about his take on it. A title like “The Art of Starting with Nothing” (orange thumbnail, above) can easily be his next Substack heading—starting out as a copywriter from nothing. Sometimes it’s that simple. Imagine someone asked you for your take on a title or headline. Think about how you would answer it, and then write.

Another thing to note, when you look at headlines daily, you’ll notice most are generic, like “How to remember everything you read.”

The top-performing titles are pretty formulaic.

Say you’re selling pencils. Look for content that’s outperforming to get a sense of what people want, and then connect it to your product.

For example: You notice videos titled “How to remember everything you read” performing well. Then, you think about how you would answer that question for your product: how do I “remember everything I read”–while talking about pencils?

Justin approaches it like this: “Notice what’s in their hand when they're using their [remembering] technique: it's a pencil. There's nothing worse than when you're writing with a pencil and the pencil keeps breaking. Well, these pencils are designed not to break. They’re also the pencils Steinbeck used.” This is how you find a topic that interests people and channel a desire onto it, even if it’s not literally about your product.

Ad concept for pencils.com by Justin Oberman.

A headline about pencils gets ignored by anyone not looking for a pencil. A headline like “He used one to write the Gettysburg Address” ties the pencil to something people already want: to write like Lincoln. Of course, they’ll need the right pencil for that.

Headlines don’t have to be about the product, but they do have to be about something the reader wants. As Justin puts it, “The headline’s got nothing to do with pencils, but it has to do with something that interests people. Because no one's going to read a thing about pencils.”

For you: Take successful efforts elsewhere and see if you can “translate” them to your work. It won’t always be one-to-one, but this is how AI can spark new ideas and creativity.

Use AI as a story library

Justin has folders for stories, ideas, projects, and clients

Before AI, Justin clipped interesting newspaper stories and ads that knocked his socks off and filed them in a folder. Today, his process is higher-tech.

He creates a Google Doc and a Claude project that links to it. Any time he comes across a useful story, he opens up the project and puts the link in and it's saved for context.

His tool of choice for organizing his copywriting research is Eden. You can set up projects in most AI tools to do the same.

Keep a separate folder for each client

Justin keeps separate folders for ideas and stories, plus one for each client. Inside each client folder are stories, strategy documents, anecdotes, transcripts, quotes, and ideas. As a bonus, the knowledge base works as a fact-check so he’s confident AI isn’t making anything up.

Any time he sees a post or video getting incredible engagement, he saves it. Then he looks for ways to write about a topic the way his client would. With their perspectives and experiences captured in a source of truth, he can point AI to the folder and ask for a draft, and that context gives his content a real voice.

The stories don’t have to come from his clients, either. He saves anything that gives his prose color.

When he needs a story for a post, instead of bothering his client, he’ll ask AI. It saves time for everyone. There’s compounding growth as he adds material over time. With clients who he’s worked with for years, all he does is ask: “Hey, is there any story that fits here?”

For you: Save your best posts, stories, and anecdotes. Then ask your AI project for additional material for your work. Examples: "Is there any story that fits here?" and "Is there anything from my storyboard I can use to introduce this newsletter?"

What’s in Justin’s project knowledge base

Write with AI. Edit with your brain.

What happens in the editing will separate you from other writers. When you write marketing, your audience really doesn't want to read what you wrote. You have to get their attention and keep them reading, and you get there by editing.

“Editing, for me, has always been 75% of writing. But now, it's even more.”

That's why Justin uses his “Copy Hunter” method. It’s a hunt for 52 different patterns.

Here’s his process:

  • Edit the body first. On Justin's first pass he looks for the standard issues: copy that doesn't flow, arguments that don't make sense, structure that isn't logical, anything obviously AI-ish, and vague sentences that lack detail. "There's a lot of moving around."
  • Write the headline last. Only after the piece is done does he figure out the subject line or first line. His selling argument can change, and the first line gets “like 90% of your effort.”
  • Run AI detectors as a gut check. When he's done and has his subject line, he runs the finished edit through three different AI detectors, including Pangram, "just to get a general gist."
  • Test subject lines against outliers. He looks for outlier posts based on the desire he's after and pulls subject lines from them. Substack lets you test five headlines and subject lines. He writes some, takes some of theirs, and puts them to the test. "Sometimes theirs wins, sometimes mine wins."
  • Assume you buried the lead. "Count seven lines down, find the line that gets to the crux of my argument, move that up to the top, probably delete the first four sentences, swap that out a little bit, and then I'm done."

As a finishing touch, drop this into your AI when you’re done: "I want to make sure that this is as good as it gets. What do you recommend?"

Client newsletter, start to finish

Some of Justin’s clients work in areas so specialized that no outside writer could come up with topics to write about. Instead, he has clients spitball ideas into AI and send him their notes. His job is to make it readable and more interesting.

One client's draft wasn't working. "It took me seven minutes to figure out what the problem was": the AI didn't have the information it needed. The fix took an hour and a half.

First, Justin called him and "started to argue with him gently about the point he was making." That drew out what the client really felt about the issue: a distinctive perspective that became the newsletter's angle. The client didn't think people would read that. Justin told him "he was right. But only if it was presented that way from the get-go."

He asked the client to have his AI rewrite the newsletter from that perspective (because only he could provide the detail needed) and let Justin do the rest.

When the new draft came in, Justin edited it and tied it to a validated topic that would immediately grab people's interest.

The result: the newsletter had the client's best open rate in years (55%), plus more than 200,000 views and 120 new

Get dangerously good at writing whether you use AI or not

It’s easy to dismiss AI as hurting your creativity but Justin believes if you embrace AI in the right places in your research and creative process, you can get more human, more creative work that beats out AI slop in the long haul.

For Justin, AI provides better context, challenges his thinking, and keeps him on the pulse of what audiences want. It exposes him to more creative jumping-off points and helps him draw connections that develop into original ideas. It’s how he writes copy that gets read.

“If you're a really good writer, there's never been a better time in the world to be one.”

What this article says