This Is How the Every Editorial Team Uses AI

TL;DR: We’re hosting a live workshop on writing with AI this Friday, co-hosted by Katie Parrott, staff writer and AI editorial lead at Every, and me. Katie will introduce her full process for writing with AI, cover why writing with AI is fundamentally different from coding with AI, and demo the tools she uses daily, including Claude projects, custom Skills, and Spiral. Ahead of the workshop, we’re sharing a deeper look at Every’s philosophy of writing with AI and our team’s workflows. Register for the event.

Was this newsletter forwarded to you? Sign up to get it in your inbox.

Today, Every is publishing our editorial guidelines. AI is woven into how we produce written and visual content for our subscribers, both as a tool to make us more efficient and as a creative partner. We want to be transparent about how that works, and we hope these guidelines can serve as a model for other AI-native publications figuring out their own approach.

The guidelines outline our mission, how AI fits into it, who we write for, and our commitment to editorial independence. But workflows are personal. Every person on our team—writer, editor, video podcast producer, social media specialist—has developed their own way of working with AI. Below, we share how.

Kate Lee, editor in chief

At Every, a draft goes through several rounds before it gets to me: a developmental edit to work out the thesis, structure, and argument, and a line edit for prose. My job is a top edit, a final pass before publication—often the first time I’m reading the piece at all. So I need to move fast and catch every single detail at once.

For subjective calls, I use Every’s writing tool Spiral. When something reads like jargon or just doesn’t sound like Every, I’ll ask for rewrite options and iterate on whichever direction I like best. My goal is always language that sounds like a specific human wrote it for a reader who’s smart but not necessarily a specialist.

For the pattern-based ones, I built a top-edit skill that screens for the patterns I catch most often: vague “this” or “that” openers without a noun to follow, unsourced quotes and data, AI tells like correlative constructions and formulaic transitions, hedging phrases, marketing speak, and sentence fragments. It’s mostly there for other editors on the team—when I’m reading a piece fresh, I catch most of this myself. If I’ve already been through a draft once, I’ll run it to make sure nothing slipped past me.

Apart from editing, I use Monologue to brain-dump an email with editorial feedback or initial notes on a research report. If I can get even a rough draft out of my head, I’m revising instead of starting from nothing.

This app that might actually make you love meetings

Meetings aren’t the problem. The challenge is everything that comes after them.

With Granola, the AI Notepad for people with back-to-back meetings, you can avoid context switching, the cognitive load of remembering what you promised… and the stress of knowing something important slipped through.

Meetings are no longer about scrambling to keep up. With Granola, they are about connection. They lead somewhere. Take notes the way you always have. Granola works in the background, turning conversations into clear summaries, action items, and next steps. Before, during, or after a meeting, you can chat with your notes to quickly understand what needs to happen, write follow-ups, or share context with others.

Download Granola for free and use it in your next meeting using the code EVERY.

Want to sponsor Every? Click here.

Eleanor Warnock, managing editor

When a draft or pitch from a writer arrives, I use AI to get a first-pass judgment—a quick read on whether the piece has potential and fits what we publish. When I have a full draft, the same skill also helps me understand whether it has the fundamental building blocks to work both as a piece of writing—such as a coherent thesis—and as an Every piece.

When I am line editing a piece for publication, AI is especially helpful for breaking down technical passages. I go paragraph by paragraph, working to make dense material accessible to a sophisticated but non-technical audience. If I’m in a rush, I will use Every’s dictation app, Monologue, to prompt as I edit in real time. As a final step in the editing process, I run a style check skill to check if the writing matches our house guide. All of the skills that we have created for the editorial team—the first-pass triage tool, AI tells skill, and style guide checker—are saved in Every’s GitHub so they can be easily updated.

Despite the perception that text-generating tools have disrupted writing, I still feel as if we’ve only scratched the surface in terms of AI permeating the messy work of editing and writing. I’m excited for when people make more tools for us.

Katie Parrott, staff writer and AI editorial lead

For at least the last 12 months, every essay I’ve written for my column has started the same way: I open my Working Overtime project in Claude and Monologue, and then I say some version of: “I have an idea about X. Can you interview me one question at a time to draw out what I think?”

I use AI throughout the drafting process, from outline to draft to revisions, but it’s far from a linear process. I might get to the outline stage and realize I’m missing a thesis. Sometimes I’ll be halfway through a draft and realize the structure is busted, and I need to go back to the outline. If traditional writing is construction—laying words brick by brick—AI-native writing feels more to me like sculpture or pottery. You get a lump of material and shape it into what you want it to become.

Often, AI-native writing is framed as prioritizing speed at the expense of rigor and craft. My own experience has been the opposite. With the grind of putting every single word after the other taken more or less off my plate, I have more mental bandwidth to think about the craft going into the piece, such as whether the introduction is compelling or the thesis is solid.

At the end of the day, though, the writing has to pass the same tests it would have before I ever heard the word “ChatGPT”: Does this writing articulate something true? Is there something here that someone else could learn from, or that could help someone feel less alone? Does the writing “sound” like me and use the kinds of words and images I would use? If the answer is yes, we’re done. If it’s not, we keep going—the machine and I—until it’s right.

Jack Cheng, contributing editor

AI tools are my relief pitchers, my bullpen. All of the pieces I edit, I read and edit on my own first. But I’ll use AI when I’m on a deadline, and I’ve been staring at a piece/section/paragraph/sentence for too long and need an outside opinion. I also run Kate’s top edit skill before I send it off to her. Claude and Spiral help generate alternate headlines and subheadings, too, though even the best ones typically require manual revision.

When it comes to my own essays, AI tools are a starting point for research alongside regular Googling. They help me identify books and articles I might want to read. I’ll sometimes use Claude in the place of a thesaurus and ask it things like—and this is an actual prompt—“Is there a word similar to ‘side eye’ that is more about suspicion than disapproval?”

Dan Shipper and I sometimes bounce ideas off of each other for his Chain of Thought column, and I’ve been refining a skill I built to identify any gaps between what he’s written about in his column, and what he’s talked about elsewhere online or internally in meetings. This skill lets Claude (or, increasingly, my OpenClaw agent, Pip) read Every’s Notion workspace, Discord server, and all 2,000-plus articles on Every’s website using both the internal and external APIs and MCPs available to our agents.

Aside from maybe the thesaurus bit, I was doing none of these things a year ago, so I imagine my workflow will continue to evolve as these tools improve.

Rachel Braun, AI & I producer

Most of my AI usage happens after the microphone turns off. Once an episode is recorded, I run it through the video and audio editing app Descript to generate a rough transcript and use its AI features to flag filler words that can be cut. Its feature Studio Sound makes sure the audio is clear, no matter where the recording took place. That transcript becomes the foundation for everything else in the production workflow.

For packaging, which is how to present the episode to make people hit play, I watch the episode to decide what would make the best introduction and highlights, then use that information to draft titles and YouTube thumbnails using ChatGPT. I’ll feed it the episode topic, guest details, and standout moments, and ask for a range of options. The best ideas usually come from going back and forth on what it gives me, rather than taking anything as is.

I also use AI tools to create shorts for social media and add closed captions with Descript or a social video editing tool like Overlap. If we are doing a livestream, we use StreamYard, which has its own AI clip-making tool that identifies the best sections and adds closed captions.

On the operational side, I’ve built a custom skill in Claude Cowork that I run each week to update our data tracking spreadsheet, pulling in performance metrics and keeping everything current without the manual copy-paste that used to eat a chunk of my week.

Six months ago, I was mostly using AI to get quick answers or speed up repetitive tasks. Now I use it more like a thought partner, especially when I’m working through title options or figuring out how to frame an episode.

Anthony Scarpulla, social media manager

Six months ago, I wrote every social media post by hand. Now, most of my work is done with Claude Code and Claude.

Inside Claude Code, I’ve built my own bespoke version of Sprout Social—a social listening tool linked to the X API that flags when people are talking on the platform about topics Every covers, so I can reply fast. It pulls from relevant past Every articles via our model context protocol (MCP), and it’s connected to Typefully, so I can draft posts quickly. For bigger strategy work, I use Claude’s desktop app to brainstorm and iterate on ideas. For creative tooling, I use Higgsfield to generate AI videos and images, and Manus for artifacts like LinkedIn slides.

When the team publishes a new article, I feed the full text into a Claude Project and ask for building blocks, not finished posts. It extracts quotable lines and proof points backing up the piece’s thesis. From there, I generate five to seven post options, a couple of thread structures, and a LinkedIn adaptation. Every batch runs through my Every style guide and the AI tells skills inside Claude—one checks brand voice, the other catches the subtle patterns that make posts feel robotic—then syncs to a Google Doc so Kate can review it before anything goes live.

The last step is mine. I read everything as a reader. If it feels like brand broadcasting instead of a friend reporting from the frontier, I kill it. I think of the role like a DJ—when AI can generate 50 tweet variations in seconds, my taste is what makes the difference between a post that sounds like Every and one that sounds like every other AI newsletter.

Kate Lee is the editor in chief and general manager at Every. You can follow her on X at @katelaurielee and on LinkedIn.

To read more essays like this, subscribe to Every, and follow us on X at @every and on LinkedIn.

We build AI tools for readers like you. Write brilliantly with Spiral. Organize files automatically with Sparkle. Deliver yourself from email with Cora. Dictate effortlessly with Monologue.

We also do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

Get paid for sharing Every with your friends. Join our referral program.

For sponsorship opportunities, reach out to [email protected].

Help us scale the only subscription you need to stay at the edge of AI. Explore open roles at Every.