Latest / Content, Briefly / How Every Builds Content as an AI-Native Company with Katie Parrott

Transcript

How Every Builds Content as an AI-Native Company with Katie Parrott · Content, Briefly

• 0:00
  [upbeat music] Hey everyone, Alex here. Let me tell you about Userp. That's U-S-E-R-P.io.
• 0:14
  If you're a growth stage or public company and SEO isn't driving enough pipeline for you yet, or maybe you've hit a growth bottleneck, you need to talk to this crew.
• 0:22
  Userp has spent 10 years helping some of the biggest SaaS, fintech, and healthcare brands in the world turn organic search into a legitimate customer acquisition engine, and now they're doing the same for AI search.
• 0:32
  We're talking companies like Monday.com, Robinhood, CrowdStrike, Wiz, Binance, and dozens more that you've heard of, from Series A all the way to publicly traded. And the results speak for themselves.
• 0:43
  They helped Monday.com go from six hundred thousand to one point two million organic visits leading into their IPO.
• 0:49
  They built hundreds of high-authority links and content assets for EarlyBird, a fintech app driving over twenty thousand acquired users before EarlyBird was acquired by Acorns.
• 0:58
  And they've helped Onboard rank for thirteen thousand more keywords in their space, tripling their organic pipeline value in a single year.
• 1:04
  What makes Userp different is they don't just publish content and hope for the best. They engineer content assets that rank and convert, backed by links on sites like HubSpot, Gartner, and G2.
• 1:14
  Here's the deal for Superpath. Mention Superpath when you book your strategy call at U-S-E-R-P.io, and they'll add five bonus high-authority link placements to your first month on top of your package. That's Userp.io.
• 1:27
  I'm Eric Doty, head of marketing at Dock and head of community at Superpath. Today, I'm sitting down with Katie Parrott, who is the staff writer and AI editorial lead at Every. I'm a super big fan of Every's newsletter.
• 1:41
  It's all about using AI for work. Katie describes it as like anyone who's an AI builder, this is the newsletter for you.
• 1:48
  It's a really cool mix of, like, editorial opinion columns, breakdowns of new models pretty much the day they come out, and just, like, really practical guides on how to use AI for work.
• 1:58
  You can check that newsletter out at every.to.
• 2:01
  So I wanted to have Katie on the show because, like, the content itself is very good, but it's also just a really well-executed editorial strategy that I really wanted to learn more about.
• 2:13
  Katie comes from a content marketing background, including a stint at Animals, which show listeners will be very familiar with.
• 2:18
  So in this show, we get into how Every runs their editorial process, how she manages to, like, stay so on top of AI, and we get into Katie's own bot-powered writing pipeline, including an agent she named Asshole that gives her work the least charitable read possible.
• 2:36
  I thought that was really funny. I also asked her top tip for any content marketer who wants to level up their AI workflow that you'll definitely want to stick around to learn.
• 2:45
  And heads up before we get started, we are hosting a Slack AMA with Katie in the Superpath Pro Slack on April 30th. So as you're listening, think of your questions for Katie.
• 2:56
  You can join the Superpath Pro Slack community for free if you're already not a member for 30 days at superpath.co. That's superpath.co. All right, let's get into it. Hi, everyone.
• 3:07
  This is Eric Doty here with Katie Parrott, who is the AI editorial lead, I believe, at Every. Katie, welcome to the show. Can you introduce yourself to our guests? Tell us a bit about your background, please. Yeah.
• 3:22
  So I am Katie Parrott, and I am a staff writer and AI editorial lead at Every. And my background is primarily in content marketing.
• 3:31
  I actually got my start, my first job when I disabused myself of the idea that I was gonna go to graduate school for German studies, was writing pitch decks for startups.
• 3:42
  And then from there, I sort of made the transition into content marketing for B2B SaaS. Kept-- Spent a couple years at Animals, and then sort of spun out on my own as a freelancer for a while.
• 3:52
  Freelancing is actually how I first got connected with Every. I co-wrote some pieces for them with the investor Lee Jin, formerly of Andreessen Horowitz. And so that's when I first gotten connected with Every.
• 4:03
  And then when I was looking for an opportunity to pick up some more work about a year and a half ago, the editor-in-chief, Kate Lee, reached out to me and said, "Hey, do you wanna come write for us?"
• 4:14
  And so the first piece I wrote was about hiring ChatGPT as my career coach, and that eventually grew into my column, Working Overtime. And now, a year and a half later, I am here doing the...
• 4:27
  still writing, writing with AI and hopefully bu- and building skills and tools that other people can use to build AI into their workflows as well. Yeah, it's super cool.
• 4:37
  There, there's so much there that I wanna dig into. Before we get too deep on the content work you're doing, can you explain Every as a company?
• 4:45
  'Cause I think, like, from my perception from the outside, so many people talk about, like, let's build a media business, and, like, if you build a media business first, then, like, that's a great content strategy, and once you've built an audience, you can do a lot with that audience.
• 4:58
  I think that was, like, a sexy thing to say a couple years ago, and what most people did is, like, they have a newsletter and a podcast, and it's not really, like, a media business.
• 5:07
  But f- what I understand about Every is that you've... it's more of, like, a subscription platform that you have, like, all these columns and you have software products.
• 5:17
  So can you tell me about, like, the business model and, like, when was it founded? Like, what are your products? Just sort of, like, give our audience the whole scope there. All right. Here, big, deep breath. Okay.
• 5:28
  [laughs] Every is the only subscription you need to stay at the edge of AI. So what we have going on is sort of an ecosystem with three main pillars. The first pillar is ideas, and that is largely the media business.
• 5:41
  That's our daily newsletter with our columns like Vibe Check, our day zero reviews of new models and tools that are coming out, source code where we write about what our software engineers are learning about building products, and that's where we develop signature concepts like compound engineering.
• 5:57
  Our co-founder and CEO, Dan Shipper, has his own column, Chain of Thought, where he writes what he's thinking.
• 6:02
  And then I have my column, Working Overtime, and then Context Window is our twice weekly digest where we sort of wrap everything up and give you little nuggets of things to think about. So that's the media.
• 6:13
  Well, and then in addition under media, the media umbrella, we do have our podcast, AI&I, and then we're increasingly getting into live streams, really experimenting with multimedia as we look for more surface areas that people are looking for information that we know we can help with.
• 6:27
  So that's the first pillar. The second pillar is our product studio. So we started life as a newsletter company. The business was co-founded by Dan Shipper and NathanBa- Bachez, Bachez?
• 6:40
  I've never actually heard him say his name out loud- [laughs]... which is really embarrassing.
• 6:44
  But they were writing their newsletters, Super Organizers and Divinations, and sort of during the pandemic just kind of teamed up and said we could start a newsletter company together.
• 6:53
  So the whole building a media company concept was come by very honestly. Mm-hmm. That's literally how the business started, and it is still very much the heart and soul of how we operate and what we do.
• 7:05
  But along the way, a little thing happened that you might have heard of called AI. Heard of it a couple times. And, uh, [laughs] yeah, hu- huge, huge, huge thing.
• 7:14
  But that really, from what I understand, I was sort of watching from the outside at this point, but that was really a watershed moment for the company because it really clicked, like not only could you use this, this technology as a tool for thought, which is something that, you know, Dan Shipper loves, but you could also build things with it.
• 7:32
  So the product pillar, which is the second pillar, really was again, really come by honestly with... It was just what the team was doing, was build- building with AI, seeing what they could build.
• 7:42
  So for example, N- Nathan spun out Lex, the AI writing tool out of Every a couple years ago. Our writing tool Spiral started as a think week project.
• 7:51
  Think week is a twice weekly off-site we do where we leave our normal work behind and just work on what's interesting to us. Same thing with Sparkle, and then we have a couple of other products.
• 8:01
  Monologue is our AI dictation tool. That, and then we have... What am I forgetting? Spiral, Sparkle. I can pull it up on your website.
• 8:11
  [laughs] Cora is our AI powered email tool, and it's really embarrassing that I would forget that one because Kieran...
• 8:17
  I work with Kieran on his pieces for Source Code, the general manager Kieran Clausen, father of compound engineering.
• 8:24
  And then our most recent product is Proof, which is an AI native markdown editor, agent native markdown editor I should say.
• 8:33
  And all of that is getting wrapped together now in our new platform, which is called Plus One, which is a hosted version of OpenClaw that comes preloaded with every approved skills, connections to our products, and just becomes like the unifying surface area where we're helping people get work done with AI, because that's ultimately what we're interested in, is helping people do their best work with AI.
• 8:56
  And that then brings us to the third pillar, which is training. So we do courses, camps and consulting. Courses are sort of standing trainings that we do on things like get started with Claw Code.
• 9:09
  Camps are more like of the moment, here's like how we're thinking about compound engineering right now. Here's how we're thinking about agent native.
• 9:16
  And then the consulting is where we work with leaders inside tech and finance companies to sort of get at them up to speed on AI and then help them transition their business into more AI native operations as well.
• 9:29
  So that's the view. That's incredible. That was actually a test, and you did pretty well. You scored an A on being able to list all of Every's product. I'm looking at your website, and yeah, you did. You got them all.
• 9:38
  Okay, so much to unpack there. So much interesting stuff happening. And one thing I want to get to as well is, like, is this all covered in one subscription? Like, the... Is that how, is that the business model?
• 9:50
  Like, you subscribe to, like, a paid version of the newsletter, and you get the software? Or how does that work? Like, how do you guys- Yes... make money?
• 9:57
  Well, this is something, like, how we articulate it and how we think about it is probably in flux just because we are sort of like, you know, sort of shifting into this agent native, you know, product ecosystem- Mm-hmm...
• 10:09
  that is, like, that is unified through this Plus One com- product.
• 10:12
  But yeah, it's ultimately a subscription-based model and, you know, you get inside the Every ecosystem and, you know, we want you to have the best tool that's for you to get the outcome that you want, whether that's a piece of content or a podcast or a product or a skill.
• 10:29
  Skills are the new IP as our- Mm... head of consulting, Natalia, likes to say. So, and skills are thought leadership is what I like to say. So we really don't care what...
• 10:40
  It's not that we don't care, but we want the right form factor for, like, the value that we wanna provide, and we want our subscribers to have access to it all. That makes sense. And who is your typical subscriber?
• 10:50
  Like, is it mostly, is it... I know marketing obviously is one of your ICPs. Like, does each of the... I think you have 12 plus columns. Like, do they each target a different ICP?
• 11:01
  Like, can you sort of paint that sort of whole umbrella picture for me? We've really coalesced around, uh, I think a pretty clear and consistent ICP, which is the AI native knowledge worker- Okay...
• 11:15
  or, like, the builder. So we talk about builders, but builder often comes with implications of, like, you have to be technical, you're writing code. Mm-hmm.
• 11:22
  And, you know, we think of builders being much more broad, you know, marketers that are building dashboards, customer success people who are, you know, doing AI native customer support.
• 11:32
  You know, we use AI across all of these departments and disciplines in our own company. Mm-hmm.
• 11:38
  And so ultimately what we're doing is taking what we learn, me from writing, Anukshi from project management, you know, Austin, our head of growth, from running growth, and we're just sharing those learnings out because we really see ourselves as a prototype of a new company model that's emerging of this AI native company.
• 11:58
  Okay, that makes sense. All right, let's get into, like, content and marketing talk. So you are the, you said staff writer, but also AI editorial lead.
• 12:08
  Can you paint us a picture of, like, what your job is and, like, what the editorial team looks like and, like, what does the marketing team look like? Can you go through that for us? Yeah.
• 12:19
  So as far as the AI editorial lead job function goes, that's something that's kind of evolved over time, and this is one of the realities of working in this AI landscape that's moving so quickly.
• 12:30
  So six months or a year ago when I started, it was really just building projects. Like, projects were a big aha moment in my personal workflows with AI, and I just...
• 12:39
  So I started by building a style guide and custom instructions for working overtime. That was working really well. I was seeing really good results from that.
• 12:48
  And eventually Dan and Kate were like, "Hey, can you take some of what you're learning and build it for the rest of the team?"
• 12:53
  So I built an AI editor project that applies some of our best practices around how we'd like content to be structured, you know, not just, like, style guide stuff like no spaces around em dashes, although we care about that very much, but things like, you know, we want a clear thesis and promise in the intro.
• 13:11
  Like, is the hook doing what it needs to do? Is the structure logical? And stuff like that.
• 13:15
  So I built that just to take-A little bit of lift off of our editing team, which we do have editors, which is one of my favorite things at Every and I'll talk about the team composition in a minute.
• 13:26
  But more recently, the AI editorial function has shifted more toward really...
• 13:31
  I mean, this is actually what it's been the whole time, is just playing around with how AI can fit into my own workflow, and then sharing out what I learn and what I think might be useful both for our team and for readers, whether that's like the AI style guide that we published a few weeks ago, or a skill to sort of check for AI tells and get the AI smell off of your writing, building various solutions like that.
• 13:55
  And one thing I should emphasize is like everybody on the Every team is a builder, so I don't necessarily own the AI editorial role the way that I used to.
• 14:04
  Like, for example, our senior editor, Jack Chang, just built a skill today for like-- or shared a skill that he and his plus-one built that is an interviewing tool that I'm really excited to try.
• 14:16
  So everyone's building stuff and sharing stuff, and I just happened to be at the right place at the right time to get the title. [chuckles] Yeah. And tell us about the team, too. Yeah, perfect. Yeah.
• 14:27
  So we are five and a half people strong now, led by our fearless editor-in-chief, Kate Lee. We then have our managing editor, Eleanor Warnock, who started in November.
• 14:38
  Jack, who I mentioned, is another senior editor who just came aboard. And then there are two staff writers, myself and Laura Entis, who's brand new. And then we share Mike Taylor with the consulting team.
• 14:49
  He's our head of tech consulting, but he also writes his column, Also True for Humans, and he participates in vibe checks.
• 14:55
  Like, we published a vibe check today on the new Cursor version that just launched, and he's in there as well. Can we actually just get right into those vibe checks?
• 15:04
  'Cause, uh, so I subscribed to the free version of your newsletter, but it's incredible, and the thing I like most is that you're always sort of on the day one testing of new AI tools that come out, like whether it's Claude Opus 4.6 or Sonnet 4.6, what- you know, whatever.
• 15:21
  Every new model that comes out, you seem to, A, already have it, seem to have it in your hands.
• 15:26
  Like, are you doing early testing of these tools before they're released to the public, or are you literally getting them like day one midnight and then writing an article? Like, how does that workflow work?
• 15:37
  Oh, good lord. No, we are not, we are not writing the kinds of vibe checks that you see on the site day of. Those are- Yeah, 'cause they're super thorough. Yeah.
• 15:45
  We're very lucky that the labs and the AI tool companies, they really seem to value our perspective and our expertise, and so we get early access to big model releases, typically anywhere from a two days to a week before the product launches.
• 16:01
  And then, you know, our value add to the ecosystem and the reason that vibe checks exist when like, you know, news companies are covering these releases too.
• 16:11
  So the why us is that these are hands-on reviews tested by the people who actually use these tools, whether it's a software engineer testing the new coding model, or it's me testing a new model's writing style.
• 16:23
  Like, it's very hands-on, very intensively sort of pressure tested, and then we're able to put those findings out when the embargo lifts.
• 16:32
  As long as all of our ducks line up in a row, we're able to get those out pretty quickly. And what does your internal process look like there? Like, I know you have your hands in the tools and you're testing.
• 16:41
  Like, y- you know, do you send it out to the team and everybody experiments with something and reports back? Like, what's the actual process of putting one of those pieces together?
• 16:49
  It is a little chaotic sometimes- [chuckles]... just because things move so quickly.
• 16:53
  We are working on some AI editorial solutions to like put a little bit more shape around how we execute vibe checks specifically, but we have a dedicated channel for every new release, and that's where we encourage people to just drop in with their findings and what they're testing.
• 17:10
  And, you know, as if I'm the writer on that vibe check, I'll go in and I'll pull out those insights and sort of synthesize them with help from my plus one, Margot, and then go in and ask follow-up questions, push people a little bit more.
• 17:23
  I'm like, "Okay, why is that annoying?" Or, you know, "What would you expect to see versus what you're seeing?" And like, "How does this change your workflow?" And so I write out that draft.
• 17:33
  It goes through, you know, editing by Eleanor or Jack, and then a top edit from Kate.
• 17:38
  And then we're also leaning on the testers themselves to go in and make sure that we're accurately reflecting their experience in that vibe check. Okay, very cool.
• 17:47
  Let's talk about just your content strategy and like content cadence generally. Like, so you have 12 plus columns from what I could, for I could see, a podcast. Like, what is... Do you have a content calendar?
• 18:03
  Is it even possible to have a content calendar with AI tools dropping suddenly?
• 18:08
  Like, how do you think about sort of the, on one side, like the content strategy overall, and then also like your calendar and your cadence as well? We do have a content calendar.
• 18:19
  It is often in motion as things change, but Eleanor, our fearless managing editor, owns the, the content calendar and sort of keeps things on rails.
• 18:28
  I will say it's funny that we're having this conversation today of all days because we just came out of an editorial team offsite for Every, where we really are just talking about, like for Q2, what we really wanna do is think about how can we rethink our newsletter for this moment of sort of agent native company building, agent native work.
• 18:48
  You know, the once daily deep dive, like highly editorial newsletter format that has been our signature for six years, you know, it's kind of reaching the edge of its usefulness in terms of best serving our audience and getting them the information they need to fulfill our promise of keeping them at the edge of AI.
• 19:07
  So what this offsite was about was just going off and thinking about what are the sort of...Atomic units of content that we would wanna see, like what are different kinds of content formats we can serve to our audience maybe packaged differently than those real, those long-form deep dives.
• 19:23
  And those long-form deep dives will still continue to live. Vibe checks aren't going anywhere, source code, working overtime, chain of thought, thesis being our, like, external voices column.
• 19:33
  Those are all, will all still be there, but we're actively thinking about, okay, in this moment when people are feeling really overwhelmed with the pace of AI change, and, you know, we now have this secondary audience of agents and LLMs crawling our content, like, what does the AI native content strategy look like in light of all of those changes that are happening around us and happening within the business as well?
• 19:58
  I wanna talk more about the editorial format specifically. So I come from an SEO background. Like, I'm more in the how-to article listicle space. I've never run, like, a regular editorial calendar.
• 20:10
  So you have some great pieces that are, like, you know, think pieces on AI.
• 20:14
  Like, you, you wrote a really great article that said something along the lines of, "I expect it to work," or, "AI is supposed to save me work, but now I feel like I'm working more."
• 20:22
  Like, uh, how do you, A, come up with those ideas on, like, such a regular cadence to be able to publish with such frequency, and then, like, what makes, like, an editorial process different than, like, let's say, like, an SEO content process?
• 20:37
  Like, what does that take on the back end? Yeah. So I'll start with the second question first.
• 20:41
  In terms of what it takes to run editorial this way, there's a term that we used to use at Animals actually that I liked a lot, which is movement first.
• 20:50
  So rather than looking outward and seeing what keywords are out there that we wanna rank for, we're really starting with our personal experience and what we're learning and what we're finding first.
• 21:00
  That's particularly true for me for working overtime.
• 21:03
  You know, I can't necessarily speak to everyone's ideation process individually, but working overtime, you know, my style guide that I have plugged into my Claude project is, you know, it says it's a first-person ethnography of being a knowledge worker in the age of AI.
• 21:16
  And really what it is is just me... If there's one thing I'm gonna do, it's put my feelings on main. [chuckles] Yeah.
• 21:21
  And so it's very sort of, it's driven both by what I'm feeling and what I'm learning as I work with these tools hands-on and just sharing what I think will be valuable to the audience, to the knowledge worker audience that I'm really talking to that is experiencing these same things.
• 21:37
  And then, you know, Dan has his process that he runs in his mysterious Dan way and, you know, the thesis, the thesis pieces obviously involve a different workflow that our editors real- like, work really closely and really intensively with these authors that we're excited to publish, these outside perspectives.
• 21:54
  And then Laura comes from a journalism background, so she's really running, treating, like, our s- our internal [chuckles] Slack as a beat, just sort of looking at what are people talking about in the studio channel where we talk about our products?
• 22:06
  What are they talking about in the design channel? And pulling all that together into a context window digest that we can send out.
• 22:12
  So it's really, it's really that, that, like, s- staying close to what we're learning, what we're seeing, and thinking about the best way that we can serve that out as content. If it hits a keyword, that's great.
• 22:25
  We're also in the habit of inventing keywords, like compound engineering and agent native architectures.
• 22:31
  Those are both signature terms that we've developed around, you know, real changes we're seeing in our actual workflows as we build these products and make this content and consult with these companies. Very interesting.
• 22:44
  What is it like doing a job where you are writing about your day-to-day work as...
• 22:50
  Like, you're kind of building the plane as you're flying it, and then also teaching somebody how to build the [chuckles] plane as you're flying. Like, that's the sense I get from you and your column.
• 22:59
  Like, h- how does that feel day to day? I actually love it because it act- it maps really closely to the two ways that I learn best, which is by doing and by writing.
• 23:08
  I always say I don't know what I think about something until I write it out. Mm.
• 23:12
  And, you know, the only way that I'm able to learn all these AI concepts is just by getting my hands on the tools themselves, like building my website through Lovable, scrapping that website and rebuilding it with Margo through, and Claude Code.
• 23:25
  So, like... And it's interest- It's funny [chuckles] because, you know, my background in content marketing, I actually don't come from much of an SEO background.
• 23:33
  I've really specialized in thought leadership, and so my job has been to be the ghost in the machine- Mm-hmm...
• 23:38
  trying to help other people share their knowledge, share their earned secrets, as I s- you know, swiped from Ben Horowitz for my thought leadership article at Animals, and now I just get to do that myself.
• 23:50
  I just get to... Like, I get to, like, do what I want thought leaders to be doing, but for my own work, and so it's actually been really energizing.
• 23:59
  It's terrifying, [chuckles] and there are times when I worry, like, "Oh, my gosh, am I scraping the bottom of the barrel of what I have to say?"
• 24:05
  But that's the great thing about AI changing at the rate that it does, there's always something new to talk about. And let's talk about that. Like, what is your actual AI tech stack right now for your day-to-day work?
• 24:16
  I know that's probably a hard question to answer, but paint the picture roughly of your architecture as it sits today. Yeah. So I've mentioned Margo. Margo is AI. She's my plus one.
• 24:28
  She lives inside Slack, and so she's kind of my chief of staff, so to speak. She manages my calendar.
• 24:37
  She does sort of prioritization and sort of deadline triage, making sure that, you know, the trains are running on time across...
• 24:45
  You know, 'cause I might be working on a working overtime column, helping Kieran with a source code piece, running a vibe check, doing a studio roundup for context window, and, like, keeping those trains running.
• 24:54
  Margo's really good at that.
• 24:56
  She's in the different channels in Slack, so she can, like, ping me if something interesting is happening that might turn into content, and she also has access to my writing agent, which brings me to my second go-to tool, which is Claude Code.
• 25:10
  Claude Code, love it. It's amazing.
• 25:13
  If you haven't tried it in, like, six months, and, like, if you tried it and bounced out of it, highly recommend going back to it because since Opus 4.5 and then subsequently Ep- Opus 4.6, it's gotten so smart and so good at it, like, understanding what you want in natural language.That I really think the, I really think the universe has busted wide open for quote-unquote, "non-technical people" in terms of what they're able to do with Claude Code.
• 25:37
  So what I was able to do was I took Kieran Claassen's Compound Engineering plugin, pulled that down the code base from the GitHub repo and just talked with Claude Code about, okay, how could I adapt this concept of like an actively learning, writing environment that learns my preferences and sort of runs through these processes?
• 25:59
  So I built my own plugin that it has sort of the complete production pipeline from brainstorming, to outlining, to drafting, to reviewing, and then finalizing. My favorite part of the process is the review stage.
• 26:14
  I have a little fleet of, I call them agents, but that's kind of charitable. They're really just skills in a trench coat, and they review my work from different angles.
• 26:23
  So I have one named Hitchcock that looks for the bomb under the table, so to speak, and what I'm looking for there is, like, is there tension and suspense in this piece that gives the reader a reason to read through?
• 26:34
  Sorkin checks if the piece walks and talks or, like, is it moving forward, like, well, or is it getting stuck somewhere? I have one named Mom who just reads- [laughs]...
• 26:43
  who reads from the perspective of someone who loves the writer very much but is slightly confused, and that sort of helps me-keep me honest to making sure content is accessible. And then I have one more.
• 26:53
  My other favorite is Asshole, which gives the content the least charitable read possible to help me sort of identify the holes in the argument, figure out what objections people might have, and see if I wanna like shore those up.
• 27:05
  So that is the... And that's an example of a tool that I built for my own workflows.
• 27:10
  I'm still kind of tinkering with it, but looking toward an eye of, okay, is this something that would be useful to externalize to other people on the Every team or even potentially as a plugin, like Compound Engineering, that people could then download and experiment with for themselves.
• 27:25
  And the workflow you built in Claude Code, is that like an app you built, or does it just happen through chat? Like, what, what does that interface look like in the content process?
• 27:35
  For a while, it was, like, literally in the terminal- Yeah... and I was like, I would go back and forth in the terminal, have it write MD markdown files to the folder that I was connected to at that time.
• 27:46
  I've mostly migrated to the Claude app. I think the app interface is pretty good, very user-friendly. Again- Mm-hmm... for people who are not terminal people, don't know how to do command.
• 27:55
  For like, again, six months ago, I had to, like, copy and paste commands and, like, try to remember the commands and everything, and it was really confusing.
• 28:01
  And now I can just say, "Here's a draft," or, or, "I have an idea for a-an article," and it will be like, "Okay, you have an idea. Let's, like, interview you to get the outputs that I want."
• 28:12
  And you're doing the writing actually, like, in the Claude Code section of the app, or are you...
• 28:18
  Okay, yeah, 'cause I've built some stuff recently in Claude Code, but I built them so that I could, like, chat with them in Claude Chat. And, like, I don't always know which is the best way to do things.
• 28:29
  Like, when I'm building the code, I'm chatting with Claude Code, but I was pulling... So I don't know, is there a reason to do your writing workflow directly with code? I will say my process is extremely chaotic.
• 28:40
  [laughs] So, like, right now- Mine too, don't worry...
• 28:42
  what, like, when by the time I get to the draft stage in my little plugin journey, I'm doing a lot of copy and pasting back and forth to, like, a Google Doc or that working overtime project that I still have in Claude that I still run to sometimes.
• 28:55
  And, like, it really just depends on, like, almost what, what mood I'm in or, like, how much of a time crunch I'm on or just, like, how my brain feels like working that day. But, you know, and that's a big misconception.
• 29:06
  This is where I get on my soapbox and talk about, you know, the misconception that writing with AI means you, like, write a prompt, say, "Write this blog post for me," hit send, and then it's a vending machine, like, give...
• 29:16
  And you just take what it gives you. The real process of writing with AI is much more chaotic than that.
• 29:22
  It's a lot of, like, back and forth, like I said, copy and pasting, pulling out, like, things that you really like and then manually tweaking the parts that don't sound like you or you think could just be a little funnier or, you know, a little bit more unique.
• 29:37
  And so it's a lot of fiddling and, you know, and not as linear as I might make it sound when I talk about the plugin, but the plugin is there with that underlying architecture, so it at least understands what part of the process I'm in and what I want to work on next as a result of being at that stage in the process.
• 29:54
  But yeah, so I like the Claude Code app. I like projects. They're, they're, they rule. [laughs] And you also flagged something. What was the name of your plus one? What did you call her? Margo. Margo.
• 30:07
  So you've flagged something that I'm starting to struggle with, in that I'm also, like, a very heavy AI user, probably not quite to the extent you are, but getting there.
• 30:18
  But the challenge it's now causing me is all this multitasking brain where, like, okay, now at Doc, I run a podcast. I do SEO.
• 30:27
  We have LinkedIn, and, like, I have AI workflows for each of these things, and I'm getting, like, a sort of top-level burnout of, like, yeah, I'm creating content at, like, this unprecedented pace for one person or a small team.
• 30:42
  But, like, the new challenge that's come up is just, like, keeping all these balls in the air, and I think that's kind of, like, the next level of AI usage is like, okay, project manage me.
• 30:52
  I need something or someone to... And, like, a Kanban board used to be enough, but that's not really possible anymore because of how many threads are open at any time. So, like, how do you even get started there?
• 31:07
  I think it's a product in Every, right? Like, the plus one? Right. Well, uh, Margo started as just a regular old OpenClaw. Sure.
• 31:15
  So for people who don't know about OpenClaw, it is an open source build your own personal agent platform that was built by Peter Steinberger, proud son of Austria, which is very exciting for me as the aforementioned failed German studies professor.
• 31:29
  [laughs] And he just built this on his own time and sort of-Unleashed it into the wild, and then if you were extremely online about two months ago, all you saw was, like, OpenClaw agents taking over the Internet, starting their own website, their own social media sites and what have you.
• 31:46
  And so Margo really started with just experimenting with OpenClaws. We ran a camp because this was a, an emerging AI capability that we wanted our audience to have access to.
• 31:55
  We invited some really smart people like Nat Eliason and Claire Vo, who had, like, been at the bleeding edge, even ahead of where we were in some cases, in some respects of building with these things.
• 32:05
  We brought them in, did an hour-and-a-half-long camp where we walked people through the process of setting up an OpenClaw, and that's when I set Margo up.
• 32:13
  And it's funny that you mention Kanban boards, because actually the first thing that I, I set up for her was just I connected her to Notion- Mm-hmm...
• 32:19
  and I built my very first, my very own [chuckles] the first time I've had a Kanban board at Every where I manage my tasks, and then I linked her to that.
• 32:29
  And now I don't really touch the Kanban board, I just talk to Margo, literally talk using Monologue, the AI dictation tool that I highly recommend checking out if you haven't.
• 32:39
  It's on Mac and iOS, so I can also use it to chat with my AI Bible study buddy, which is very fun for me. But I'll just talk to Margo and, yeah, she runs things.
• 32:50
  You can set up, you know, cron jobs, like chronological jobs to run at certain times. So, like, you know, I might want a midday check-in for her to say like, "Hey, you're supposed to get X, Y, Z done today.
• 33:00
  Like, have you done, you know, A, B, C under that task?" So there's a lot that you can do with agents. It does take a little bit of elbow grease to set up the systems and, like, set the rules that you want, you know.
• 33:14
  And part of our value proposition, or part of our offering with the Plus One is that w-we handle some of that setup for you and give you a pre- you know, a preloaded cl- you know, claw basically with all these skills, but you're still gonna wanna personalize it on top of that, right?
• 33:29
  You're still gonna wanna, like, teach it the skills that you want it to have that are relevant to your discipline, and that's the challenge, but it's also the opportunity because, again, you can do so much through natural language.
• 33:41
  Just saying like, "Hey, I want you to learn this skill," p- like, link the GitHub, you know, repo, and it will go learn it, and then you tell it, "Okay, save that information so that you have that access to that skill perpetually going forward."
• 33:55
  So for example, I taught Margo, like, "Any time we're producing a document, I want you to produce a Proof document," which is...
• 34:02
  'Cause Proof is an AI, an agent native markdown tool, and it's just the g- a great surface for us to collaborate on.
• 34:09
  I'm [chuckles] currently a little stuck trying to get her to learn how to format screenshots because I hate formatting screenshots.
• 34:17
  You know, having to go into shots.so and, like, get the right background to make our designer happy and set the right, you know, form factor.
• 34:24
  And so I'm trying to get Margo to learn how to do that because really what I want her to help me with most is what our COO, Brandon Gal, calls the paper cuts, you know, the really small annoyances that are like, but, like, they're small but they're repetitive, right?
• 34:38
  And so I'm always on the lookout for those paper cuts that Margo can help me solve for, and that's really what I use her for.
• 34:44
  And then for, like, more extended work, you know, in, like, real drafting is when I go to Claude Code into my projects. Perfect. Okay, I have a couple more questions. How do you keep up with all this? I am...
• 34:58
  You know, honestly, I'm in the Slack channels, man.
• 35:01
  Like, we've got an Everyone channel that's kind of, it's a little bit of a fire hose of, like, news and what's, like, what's the latest, but it's really good signal because it's coming from a team that I trust, that I know is really at the edge of their discipline.
• 35:15
  And so if they flag something and say, "Hey, this is worth checking out," like, I know that it's worth checking out. And this is just the Every team? Yes, yes. It's our Every... We're now in Slack.
• 35:25
  We used to be in Discord, and we moved, and it's the best thing that's ever happened to me. But so we have the channels, and then I'm extremely online in terms of, like, X and LinkedIn.
• 35:35
  Yeah, where do, where are you online? Yeah, I'm, I'm on X still for my sins and LinkedIn as well.
• 35:41
  And then, you know, like, I have, like, certain influencers that I follow on there that, like, I really care about their perspective. Any come to mind as someone who wants...
• 35:51
  I mean, you're like my number one source for AI news probably, so I'm getting the source of my source. I mean, there's some names that come to mind. Claire, who I mentioned, she has her own podcast, How I AI.
• 36:03
  I really value her. Like, she's much more of an operator than I am. Mm-hmm.
• 36:07
  Like, she's the founder of a company called ChatPRD, but she does a really good job of, like, getting down to brass tacks of, like, what is useful and why, and I think she does a good job of making the information accessible.
• 36:19
  For more conceptual inspiration, Ethan Mollick, who is a professor and researcher, you know, he was interviewed in The Economist recently talking about keeping AI weird and, like, that's a framing that I find really interesting and like, okay, like, how do we...
• 36:33
  Like, you know, we can't think of this technology as being too domestic, right?
• 36:37
  It's a wild beast, and we need to let it be what it is and, like, be mindful of the, of both the positive and negative externalities of that. So he's really smart. Those are a couple of the k-...
• 36:47
  That's the kind of thing that I'm looking for. Yeah. A lot of the time it's less tactical, although I do, like, you know, I pay attention when... Oh, what's his name?
• 36:55
  He posted, like, a whole skill, like a whole set of skills for doing various functions within marketing. I was like, "Oh, I should, like, yoink that and twist it- [chuckles]... for my...
• 37:04
  Do a yoink and twist for my own workflows." The old yoink and twist. Yes. [laughs] Okay, we've probably lost this listener at this point, but if I'm an early-ish AI user, like I'm a light...
• 37:18
  You know, I've used ChatGPT or Claude for some basic stuff, but, you know, I'm a classic content lead, let's say, that's really trying to, like, up my AI usage.
• 37:28
  You've talked about Claude Code, but, like, where are some easy places to start? Is it those paper cuts, or is it, like, some part of the content workflow?
• 37:37
  Like, how would you tell people who are at the, like, 101 or 201 level to- Mm-hmm...
• 37:43
  bring their work to that next level with AI?This might be a weird place to start answering this question, but it is my favorite hack for a whole range of problems, is just asking the AI to interview you. Mm.
• 37:55
  And that works, that can work from the top down.
• 37:57
  That can work with help me identify where you can be helpful to me, help me articulate my particular preference of how to do this task and turn that into, and package that into a skill so that this work gets executed like according to those guidelines reliably every time.
• 38:14
  You know, interview me to find out what I think for a- an article that I wanna write.
• 38:18
  Like, I really think the more you can like let AI help you, help AI help you, like it can be really fruitful and really give you, you know, something better than just here's how this X person that I admire does Y, but like based on how you already work, here's how AI can be helpful to you.
• 38:40
  That's such a good tip. That has been such an unlock for me in building things with Claude Code, for example. It's I start in Claude Chat and say, "I'm thinking of a project that does this. Interview me about it.
• 38:52
  Ask me about all the pieces, what tools I wanna connect together," and, you know, I'll answer 15 questions, and then it goes and spits out like a 2,000-word prompt that I can th- then give to Claude Code and then build something incredible.
• 39:05
  But if I had done this from my own brain from the beginning, I would've forgotten 50 things. I would've used up all my credits before I even built the [chuckles] thing that I wanted to build.
• 39:14
  So yeah, that like have the AI interview you is such an incredible tip, and you can basically start anything with that advice, so that's brilliant. Let's get back to content.
• 39:24
  Like are there any pieces of content for Every that you've done that have performed really well or like in the last year or so? Yeah.
• 39:32
  So the content that we're really seeing have the most success is those pieces that I talked about that are sort of like our IP. Like we originated the concept.
• 39:41
  We found something in our workflows or looked at the technologies that we've been interacting with and sort of built a framework around that idea that we own.
• 39:51
  So compound engineering, you know, if you look that keyword, we're gonna be at the top ranking for that. We're gonna... If you switch over to the AI overview, we're gonna be referred to there.
• 40:01
  Same with AI native architecture.
• 40:03
  And then related but also kind of s- side by side with that, like this is my like where my content marketing brain really goes wild because what we're seeing a lot of success with is guides. Mm-hmm.
• 40:14
  And like if you were in the content marketing space 10 years ago like I was, you remember the time of the ultimate guide- Oh, yeah...
• 40:21
  when we were all doing the skyscraper technique, trying to one-up each other to create, make our guide even more useful, and that...
• 40:27
  Um, we've just seen that content really take off because ultimately I think what people are looking for is action steps. They wanna be able to get an outcome.
• 40:35
  And so in a weird way, we're coming back to basics in the content marketing world of can you build something that whether that reader is a human or an agent reads it, they'll come away with the ability to execute something at a really high level.
• 40:51
  So that's our guide strategy, and we've seen like huge returns from that. So if I had one piece of advice for the content marketing world, well, kind of a twofold piece of advice, is like look for those earned secrets.
• 41:05
  Look for the genuinely net new knowledge that you can add to the conversation, and build guides around it and put those out and say, "Here's... Like we know you have this problem.
• 41:14
  Here's our solution to it, and here's how you can follow those steps as well." Yeah, that's a good tip.
• 41:19
  I think recently you shared a guide on it was like making your own writing skill, like in your own style of voice, and I think it was an interview prompt or you had the prompt in the guide itself.
• 41:30
  And so I was able to, you know, copy-paste that.
• 41:33
  I think you even had a nice like copy module built into the webpage, that it was just like, "Copy this," and like so I think taking that like next level of saving your reader time as well, right?
• 41:45
  Like anyone who's messing around with AI a lot has wasted so much time, like where each of us is reinventing the wheel ourselves almost every time we go to do a task, which is kind of weird.
• 41:56
  Like w- using Kanban boards as an example, like Asana already built the board, and we all just go in and use it, and like they've already thought things through, versus so many of us when we're using AI are like building something from scratch every time.
• 42:09
  And, uh, so much of like being helpful in the world of guiding people with AI is like, "Hey, here's like 10 steps that I already figured out for you. [laughs] Here you go. Don't spend your whole afternoon on this.
• 42:20
  I, I've already figured it out." So I think that's what I appreciate from like the guide model where it's not just like 10,000 words of advice.
• 42:27
  It's also, you know, here's some shortcuts that we've learned that will help you. So it's truly valuable in that sense. Okay. Finally, where can people follow you and your company's work? Yeah.
• 42:38
  So you can find Every at every.to. Our X handle, we actually have the @every X handle, so that's- Nice... lovely. And then for me personally, my website is katieparrott.com.
• 42:51
  I have a running change log there where I update the things I've published, things I've learned, things I've built. I'm also on X, @kplike the bird, and LinkedIn, Katie Parrott.
• 43:02
  And Katie will be doing an AMA in the Superpath, Superpath Pro Slack on April 30th.
• 43:09
  I am 95% sure this episode will come out before then, so we'll make sure people give this episode a listen and then come to the Superpath Slack with great questions for you. So Katie, thank you so much for doing this.
• 43:21
  You were an awesome guest, and hope to have you back on again time sometime soon. Yeah. Thanks for having me. This was super fun. [outro music]