Ghostwriter David Johnson-Igra has spent years crafting the voice of tech executives, supporting leaders at notable clients such as Amazon, a16z, Meta, GitHub, and OpenAI. Then suddenly in April 2025, his business completely disappeared. Every one of his clients let him go within a few weeks.
While he can’t definitively pin it on the release of Anthropic’s Claude 3 Opus model and acknowledges general budget tightening in the industry, he finds the timing suspect: the model was released just a few weeks prior and was widely hailed as a giant leap. Particularly in the tech circles that make up his client base, discourse around using AI for writing and to produce blogs and LinkedIn posts faster, cheaper, and with more autonomy had already been growing. It felt like a pivotal moment, and it lit a fire under him to reinvent his business for the new AI world.
“I was like, maybe I need to rethink what I’m doing and how much I’m doing with AI for content. It started a whole process of really starting to learn how to use Claude and ask, ‘what can the future look like for me?’” Johnson-Igra told Fortune.
As he tried to figure out his next move, the narrative shifted from enthusiasm about using AI for writing to concerns about “AI slop.” New clients started reaching out, but he didn’t offer them his same old services. After breaking down every part of his process to see how AI could improve not only how he works, but also what he offers, he pivoted. Now, rather than deliverables, he’s offering AI-driven content systems — and it’s rewiring everything about his business.
The “second brain”
Ghostwriters commonly create a guide for every executive they work with, noting characteristics of their tone, voice, and how they’d speak on different topics. In his revamped business, Johnson-Igra is transforming the classic voice guide — previously just a reference for the ghostwriters themselves — from a static document into an AI-powered system that continuously learns, makes connections, and surfaces insights. And he’s selling that system to the client.
Johnson-Igra creates what he calls a “second brain” system for each executive by combining a knowledge graph, which organizes data points in a way that maps the relationships between them, with an LLM of their choice. He inputs all the information relevant to the executive’s writing — interviews they’ve given, past content, performance metrics on that content, notes on what they want to say in their writing, etc — to form the knowledge graph. Then when it’s time to write a new piece of content, he can use the system to get strategic inspiration and get the draft going.
For example, if an executive wants to write about a certain topic, this system can pull up every mention of that topic in past content, surface how they spoke about it, and also how the post performed. Then, the model can create the first draft, which he then edits. “It’s not just about efficiency,” says Johnson-Igra. The benefit, as he sees it, is that these AI tools draw connections deeper than he might’ve noticed on his own.
Overall, this amounts to a total shift in his offerings. Instead of selling content deliverables, he’s selling a custom system with an expert running it. The client owns the system, but Johnson-Igra still does the work and applies his expertise. The knowledge graph software he uses, Obsidian, is also model-agnostic, meaning it can be transferred over to use with a different model at any time, preventing the company from ever being locked in to one AI provider.
“Now those outputs are a means to an end,” he said. “The end is the system.”
Are we a tech business now?
Johnson-Igra’s new services go beyond “digital brains” and also include personalized software tools designed to enhance marketers’ work.
For one client, for example, he created a Python script to generate a content performance report for an executive’s LinkedIn, analyzing upwards of 4,000 rows of spreadsheet data. He also conducted a competitor audit for the same executive, for which he culled 1,000 pieces of content across 10 competitors and generated a heatmap visualizing the topics the leaders share.
Currently, he’s creating an MCP—a connector that allows an LLM to communicate with external tools or services—for a VC firm and one of its portfolio companies to help streamline the startup’s brand messaging. Specifically, it will instruct Claude to use specific messaging language to describe the company’s market opportunity, challenge, and target customer that both the VC firm and the portfolio company had previously agreed upon.
All of it would’ve been impossible for him not too long ago, but he’s staffing his revitalized business accordingly. While previously, Johnson-Igra hired writers to help him scale by increasing how much writing work he could take on, he’s now tapping technical help, such as a systems engineer, to broaden what he can offer. He’s also diving head-first into these tools himself, and that’s what he’s crediting his second act to so far.
“I don’t know that much,” he said. “And the only advantage that I have right now is that I keep trying to learn more.”
