The Most Valuable Step in Writing with AI Isn't the Prompt—It's Turning Editing Feedback into Compound Interest
Katie takes AI from career coach to writing plugin: first she loads the model with context, then she saves every round of editing feedback back into the system to compound. For her, AI's value isn't just efficiency—it's also a support system that steadies her emotions and helps her push through execution hurdles.
The argument · tap a timestamp to hear it
Asking AI to coach is about being challenged, not being replaced
After being laid off from an editing and ghostwriting role at a crypto company, Katie tried agency, client-side, and freelance work, yet still felt her own ideas had hit a ceiling. Hiring a real career coach was too expensive—she estimated around $150/hour—so she switched to ChatGPT at $20/month. She didn't ask it to think for her; instead, she used it as an externalized thinking partner: she explained her situation, let it probe and push back, especially pulling her out of herself when she tended to catastrophize. She later called this a "thinking tool." She even used it to decide whether to accept Every's column invitation—ChatGPT advised her to try it, and she's been there ever since.
— Katie ParrottDelivering five channels solo relies on context, not typing speed
In early 2025, Katie severely overcommitted as a freelancer: within a two-week deadline, she had to deliver 8 blog posts, 3 ebooks, 24 LinkedIn posts, 24 X posts, and 16 Instagram posts. She managed it not by writing faster, but through what she later realized is called context engineering: first writing out the company's brand messaging, product details, audience, and differentiators line by line, then letting the model generate. Building that context upfront was laborious, but once the foundation was set, subsequent production speed became very fast. She admits she repeated the overcommitment six months later—this method didn't cure her overcommitting habit.
— Katie ParrottWithout foundational information, style constraints spin their wheels
On what to put in a project first, Katie is clear: not starting with "use these words, avoid those" or "keep reading level low," but first giving the model a base profile of who the reader is, what their pain points are, how the product addresses them, and who the competitors and differentiators are. She says this is her instinct from writing style guides in content marketing, just now fed to AI. Only after the foundational material is in place do constraints on tone, wording, and syntax come. Her take on "style": wording issues will always nag, but without foundational information, style constraints are just spinning their wheels.
— Katie ParrottThe model is the kitchen; fresh ingredients must be brought in by humans
The "data" she means isn't generic background, but first-hand material that postdates the model's knowledge cutoff or exists only in the real physical world: the company's own research, third-party reports, specific personal experiences. If you only ask the model to write a "style guide" from what it already knows, the output is commoditized information. So she assigns humans the "last mile": bringing new material the model doesn't have into the system. Her metaphor: the model is the kitchen, the process is chopping and cooking, but fresh ingredients must be carried in by a person—the quality and freshness of inputs ultimately determine whether a piece is worth reading.
— Katie ParrottAI as supportive technology is as powerful as as a production tool
Katie shared a layer she rarely discusses publicly: she has bipolar disorder, experiencing high-energy periods and facing very long lows. For her, AI's value isn't just "high output"—it's also reducing friction in daily life, making it easier for her to function as a normal person. She had put off booking a family doctor for three years, then had Codex automatically find nearby doctors who accepted her insurance and new patients, and book an appointment. Her inbox is also pre-filtered by automation to surface only emails that truly need her reply, rather than her digging through subscription notifications daily for any "bombs." Her own phrasing: AI as supportive technology is just as powerful as productive technology.
— Katie ParrottA career coach isn't a single prompt; it's a project
That simple "help me as a career coach" prompt from back then has now become a standalone project in Codex: it contains her own profile and role positioning, Every's brand positioning intel, article performance data exported from the CMS, a validation folder collecting reader praise, and Q2/Q3 OKRs. Much of that material wasn't manually organized: she learned Alex Duffy's method of having AI interview her, drawing out her thoughts and organizing them into documents. She now increasingly talks to it aloud in "self-talk" mode, having it prioritize, manage projects, and maintain a kanban board. She's even forgotten what's on the board, only asking when she needs a deadline for a deliverable.
— Katie ParrottWriting should be like engineering: fix each issue only once
Compound Writing started by forking Kieran's Compound Engineering: in engineering, a system should remember every piece of feedback, fix each issue only once, and automatically do it right next time—writing should work the same. She mapped the engineering phases of brainstorming, planning, working, and reviewing onto writing's brainstorming, outlining, drafting, and reviewing, and designed three types of review: substantive edit for overall structure and argument, line edit for sentence-by-sentence polish, and a final publication-ready check. She no longer opens web chat windows; instead, she runs this system on Claude or ChatGPT's desktop apps, with style guides, sample articles, and other context inside, so the human only needs to bring in fresh information. The plugin was already downloadable from Every's GitHub at the time of recording.
— Katie ParrottThis plugin is a gym, not a ghostwriter
She calls the plugin a gym, aiming to make the user stronger rather than have AI ghostwrite. She breaks down favorite authors into invocable reviewer skills: Vonnegut on "start as close to the end as possible, every sentence must earn its place, give the reader a character to root for"; Hitchcock on the "bomb under the table"—if you let the audience see the countdown, a second of shock becomes five minutes of suspense. She builds these frameworks directly into the editing steps, so after writing you can invoke Vonnegut, Hitchcock, Sorkin, or Sedaris to review your draft. She believes non-professional writers benefit even more, because they don't have the baggage of "I used to write it better this way" standing in the way.
— Katie ParrottIn their own words · checked verbatim
Clarity, it turns out, doesn't arrive gift wrapped from a digital assistant or even a human coach. It's something I had to dig out for myself, question by question and prompt by prompt.
Natalia Quintero5:13
But when you have that environment set up, you can just kind of run and like run fast.
Katie Parrott10:28
So the data is the unique insight, the data point that the model doesn't have access to yet, the personal experience, all of those things that are working are ultimately going to make the writing unique.
Katie Parrott17:45
AI is just as powerful as a supportive technology.
Katie Parrott22:50
So compounding is the idea that every piece of feedback that you give to AI should feed back into the system to improve the next output.
Katie Parrott32:09
if there's a bomb under the table and it explodes, that's a one moment of surprise. But if you have people sitting around the table and you show the audience the bomb under the table and say this bomb will go off in five minutes, now you have five minutes of suspense.
Katie Parrott40:19
what I want to challenge the AI community with is thinking about how we can expand those kinds of opportunities to more people and more kinds of people in more different places.
Katie Parrott46:32
Figures
| Real career coach cost vs. ChatGPT subscription at the time | about $150/hour vs. $20/month | 4:12 |
| How long she delayed booking a family doctor | 3 years | 23:53 |
Glossary
- Compound Engineering
- An approach where AI remembers every piece of feedback, so each correction is given only once and automatically applied thereafter, making output continuously better.
- Compound Writing
- A plugin that transplants compound engineering to the writing process, accumulating feedback on style, structure, and editing.
- Context engineering
- The practice of loading background information—brand, audience, competitors—into the system before asking the model to write, setting boundaries for the AI.
- Supportive technology
- AI that doesn't produce content but reduces friction in life and execution, making it easier for a person to get things done.
- MCP
- Model Context Protocol, a common interface standard connecting models to external data and applications; Katie uses it to let small apps interact with her tools.
How to listen
Journalists and editors producing content, WeChat articles, or long-form pieces, plus independent developers building AI as a personal productivity system—especially those stuck with homogeneous AI output and looking to establish their own workflow.
The preview before 1:04 and the sponsor read at 21:48 can be skipped; the main content runs from 8:26 and stays valuable to the end.