She Fired a $4,000-a-Month Content Team - Two AI Tools Replaced Them

She Fired a $4,000-a-Month Content Team.

Two AI Tools Replaced Them

Carrie used to pay a content team $4,000 a month to help her run her newsletter. Now she pays for two AI subscriptions and gets the work done before lunch.

Writing on her Substack publication Thrive with Carrie, the six-figure solo newsletter operator detailed the workflow in a post titled My AI Newsletter Production Line (Under 60 Minutes). The headline number is the part that made it travel. The structure underneath it is the part that matters.

Her stack is short: Perplexity for research, Claude Sonnet for drafting and repurposing, and her own voice — captured by phone — for everything in between.

Step one happens before the AI shows up

Carrie's process doesn't begin with a prompt. It begins with her inbox.

Before opening any AI tool, she mines her own emails, direct messages and reader comments looking for the exact phrasing her audience uses when they describe what they're struggling with. The friction her subscribers name out loud is the friction she turns into newsletter topics.

Only after she has the topic does Perplexity enter the picture. She asks the search-grounded model to surface the five biggest frustrations her audience has around the chosen subject — using their own language as the seed.

It's a small reordering with a large consequence. The AI never picks her topics. The AI sharpens her angle on topics she already knows resonate.

Voice-to-text is the secret middle step

The next move is where the mobile workflow shows up. Carrie doesn't write the body of her newsletter at a keyboard. She talks.

In the same post, she described handing Claude three inputs: Perplexity's research, her outline and "voice-to-text observations and stories" she's captured on her phone. The model fuses them into a draft that already sounds like her — because the raw material already sounded like her.

The shift matters. Most AI newsletter tutorials start with a prompt and end with a generic article. Carrie's starts with her own voice — talked into the phone, transcribed, and dropped into the model alongside the research. The output never has to learn her voice from scratch.

Repurposing is one prompt away

Once the newsletter is drafted, Carrie repurposes. She pastes the finished piece back into Claude and types a single instruction: "Turn this into 3 Substack Notes hooks and one LinkedIn post in [describe your voice]."

That's the entire repurposing pipeline. No separate social-media tool. No second writer. One model, one prompt, three additional pieces of content.

The numbers she's reporting

The receipts Carrie shared in her post are the kind that make other creators screenshot:

Her newsletter writing time dropped from 4.5 hours to 2 hours. Open rates climbed 23 percent in six weeks. Engagement doubled. And the editorial team she used to pay $4,000 a month for is gone.

"It still sounded like me," she wrote, which is the line every solo creator considering AI assistance is privately worried about.

Why the two-tool stack works

Carrie's decision to combine Perplexity and Claude — rather than relying on either alone — is deliberate. She uses Perplexity because it returns citations, which lets her verify claims and link sources inside the newsletter itself. She uses Claude for drafting because, in her experience, it follows voice instructions more cleanly than other models when she pastes in her own talked-out fragments.

She isn't running a complex automation. There's no Zapier, no Make.com, no scheduled scenario. The whole production line is two browser tabs, a voice-to-text app on her phone and a published Substack post at the end.

The bigger pattern

What Carrie's setup demonstrates — and what her open rates appear to confirm — is that AI assistance in 2026 isn't about replacing the creator's voice. It's about reorganizing the bottleneck.

For her, that bottleneck used to be the four-and-a-half-hour writing block. Now it's the inbox-mining step at the start, which she still does manually, and the editorial pass at the end, which she also still does manually. The two-hour middle, the part that used to grind, is now mostly Claude.

She's still writing every newsletter. She's just no longer typing every newsletter.

The implication for other newsletter operators

The post on Thrive with Carrie has circulated through Substack's newsletter-operator community for a simple reason: the workflow doesn't require a developer, a paid automation stack or a custom GPT. It requires two tools, a phone for voice capture and a willingness to let the AI handle the cleanup rather than the thinking.

Whether other creators can replicate Carrie's open-rate gains is its own question. What they can replicate, today, is the production line. The receipts are right there in her post.

Source: Carrie, My AI Newsletter Production Line (Under 60 Minutes) (also published as I Replaced a $4K Content Team With 2 AI Tools. Here's the Exact Workflow.), Thrive with Carrie, Substack. https://thrivewithcarrie.substack.com/p/ai-content-production-line-newsletter-creators