In this episode, Nathan Baschez offers a refreshing perspective on AI writing tools. He explains we're still at "the very bottom of the S-curve of adoption" for writers embracing AI. Meanwhile, programmers have rapidly integrated these tools into their workflows. This gap represents both challenge and opportunity.
Instead of getting stuck in the typical AI debate ("Is it a threat to writers or just a shortcut?"), Nathan has a different take. He argues that neither outright rejection of AI nor complete surrender to it serves writers well. Instead, he shares a compelling framework: think of AI as a personal trainer for your writing process rather than a replacement.
"I'm not trying to avoid writing by using AI in my writing process," Baschez explains. "I'm trying to have a more satisfying writing process. Sure, maybe a little bit faster in some ways, a lot less agony. But in the end, I want to come up with something that I feel really proud of personally."
This middle path offers content teams a nuanced approach that increases creativity while maintaining the human authenticity readers want.
Nathan Baschez is the founder and CEO of Lex, a tool that reimagines writing for the AI era. His unique perspective comes from a career spent in content, media, and tech. He was previously VP of Product at Substack, Head of Product at Gimlet Media, and co-founder/president of Every, a business and technology publication bundle.
In summer 2022 — before ChatGPT existed and AI discussions centered mostly around image generation — Baschez started building Lex as a side project born from frustration with Google Docs. "I just felt this steep drop off in sort of quality or thoughtfulness of the tool," he explains about transitioning from coding tools to writing platforms.
What started as a nights-and-weekends project with a simple AI feature called "plus plus plus" (allowing writers to generate a paragraph when stuck) exploded with 25,000 signups on the first day of release. Two years later, Lex has spun out from Every into an independent company that's approaching profitability while following the sustainable growth model of respected companies like GitHub, Linear, and Notion.
As you'll hear in this conversation, Nathan is one of the most thoughtful ambassadors for a future where writers benefit from AI rather than fear it. His balanced perspective — neither rejecting nor surrendering to AI — makes him exactly the kind of pioneer we might actually trust to guide us through this transformation.
Throughout our conversation, Nathan shared insights that challenge conventional thinking about AI and writing while offering practical ideas for content teams.
Many writers still haven't integrated AI into their daily workflow, highlighting a major opportunity for content teams. While programmers have rapidly embraced AI, writers have been slower to adopt these tools. Nathan identifies both conceptual obstacles (ethics, fear of losing authenticity) and practical challenges (understanding use cases) that explain this gap.
Instead of viewing AI as a way to avoid writing, Nathan frames it as an enhancement to the creative process. This mindset shift is helpful for content teams who fear that embracing AI means abandoning craft.
In discussing his experiences at Gimlet and Substack, Nathan shares what he learned about creating extraordinary products and businesses. He emphasizes that success isn't mysterious — it comes from consistent effort, caring deeply, and persistence.
Nathan compares using AI to hiring different types of assistants. You could hire AI to completely take over a task ("Anytime you have to write anything, AI, write it for me.") Alternatively, you could use it more like a personal trainer ("Here's my writing, critique it.") This framework helps content teams be more intentional about which types of tasks to use AI for.
Nathan shares a framework for soliciting feedback on your writing that works well with human editors and AI assistants: the ABCD approach — Awesome, Boring, Confusing, and Didn't believe. What makes this particularly valuable for content teams is how well AI performs this assessment.
With programming, success is binary and objective — either your code works or it doesn't. With writing, success is subjective and contextual. It depends on how human readers respond and understand your text. This difference makes AI adoption trickier for writers, as the pros and cons of involving AI in your process are less clear.
Nathan shares a trick from his writing workflow that he's relied on more since using AI. He begins with a "brain dump." Not a formal outline but an exploration of "goals," "main things I want to hit on," and origins of the idea.
This free-writing approach "helps unpack my thoughts" and clarifies what he's writing about. At the same time, this input provides a great starting point for the AI.
Hello... is there anybody out there creating real value with AI?
The AI conversation in content marketing has become deafening — skeptics shouting from one side, shallow tips from enthusiasts on the other. But somewhere in this noise, there must be pioneers who've actually figured something out, right?
We've gone on a search for the real pioneers — the ones who've ventured beyond the hype to succeed (or fail) spectacularly. Through their hard-won insights, we'll discover if there's actually something of value hiding in the noise, or if we're all just shouting into the void.
Check out other episodes in the season here
Art of Accomplishment (00:02:04): A podcast/newsletter by Joe Hudson, former venture capitalist turned coach, mentioned by Nathan as content he's been consuming lately.
Lex.page (00:35:03): Nathan's AI writing tool that looks like Google Docs "but on steroids and cleaner."
Cursor (00:08:30): An AI-powered code editor that Nathan mentions using instead of VS Code.
Nathan's career journey (00:06:14-00:09:00): Before founding Lex, Nathan was president and co-founder of Every (a publication bundle), VP of Product at Substack, and Head of Product at Gimlet Media.
Alex Blumberg's media projects (00:06:14-00:07:00): Nathan mentions Blumberg's work at This American Life, Planet Money, and especially the Startup podcast at Gimlet as some of his favorite media and influences.
Zapier, GitHub, Linear, and Notion (00:14:00): Companies Nathan mentions as models for sustainable growth.
Lex’s Annual Letter (00:14:39): Nathan's thoughtful exploration of the AI adoption gap between writers and programmers. MUST READ if you enjoyed the episode.
Follow Nathan and Lex: @nbashaw , @lexdotpage, or email hello@lex.page
[00:00:00] Nathan: We're at the very, very bottom of like the S curve of adoption with writers and AI. You know, like most writers, maybe they use Grammarly. Maybe they pop over some questions or chat GBT like every once in a while, but it's not like they've really adopted it as a part of their writing workflow in any real deep way.
[00:00:16] Ty Magnin: This season on the Animals Podcast, we're focused entirely on AI content use cases. We're bringing you on a search to meet the AI pioneers. Those venturing beyond the hype to succeed or fail spectacularly. Today we're hosting an interesting conversation with Nathan, the CEO at Lex. Nathan has spent his career at companies like Gimlet Media, at Substack running product, and has since gone on to found Lex.
page or Lex, which is an AI writing application that looks a lot like Google Doc, but like on steroids and even cleaner. Some of the insights that Nathan shares today are around. What is blocking AI usage for content writers, some of the emotional blocks as well as like the tangible technical blocks. He talks about [00:01:00] his own workflow using Lex and using AI.
You might get a few tips of frameworks for your own usage of AI or just generally when you're editing your own work. So I, I hope you enjoy this conversation with Nathan. Did you know Animals Now offers a podcast service? We're taking over your audience's earbuds, reaching them during their commutes, their workouts, or when they're doing chores around the house.
From show strategy to editing and distribution, animals can handle your podcast for you with that same originality. Audience first approach that we bring to all of our content. Every podcast episode can become fuel for your broader content program. You can mine your podcast for ideas for articles, social posts, and other kinds of content assets, helping you create more high quality work in less time.
Ready to start a podcast worth listening to? Head over to animals. co book a call with us, and we'll start talking about your podcasting goals. Welcome, Nathan. So excited to have you on the animals podcast. One question we ask to all of our [00:02:00] guests is what content have you been consuming lately?
[00:02:04] Nathan: The deepest kind of, kind of most important content.
There's always Twitter slop, you know, or X or whatever. Like there's, there's all that of course, but the stuff I've been loving lately is, uh, have you heard of art of accomplishment? This guy, Joe Hudson has like a podcast newsletter, stuff like that. Have you heard of this?
[00:02:22] Ty Magnin: No, I haven't.
[00:02:22] Nathan: It's really good.
It's basically this guy used to be a venture capitalist and then he became a coach and, you know, coach slash kind of like therapist. Like sure. Someone, a friend of mine recommended his stuff to me. I saw he, he tweeted, like, I, my wife and I like never punished our children. And anytime I caught myself trying to shame them, I would apologize.
And like, basically it worked out really well. And so I replied, would read a 5k words on this basically, because I've got a daughter who's like two and a half now. And we're kind of getting to that, like, Okay, like how do we sort of do discipline basically, like the way my dad and mom did it was, I thought pretty good, but you know, I was [00:03:00] like, I'm just curious about this kind of stuff.
So it's kind of all these angles that his stuff is about. But the thing that's been the most interesting to me is how so much of getting in our own way just comes from resisting actually trying to feel emotions. And when you kind of welcome those emotions in like helplessness, fear or whatever, it's like you can kind of learn to almost savor them in a way like, ah, this is fear.
Like. You know, we do it when we're listening to music, like sad music. It's like something not bad about it. You're welcoming the sadness or like a scary movie. Like there's something not bad about it. And then if you can kind of learn to bring a little bit of that into your own, like, you know, hard decisions you have to make or whatever, it makes the whole thing a lot more enjoyable in a weird way, even the, even the sort of bad emotions or whatever.
So I've been loving that.
[00:03:42] Ty Magnin: Nice. I dig it. That carries over, I imagine, into your leadership at Lex, leader as a father, uh, and then elsewhere in your relationships.
[00:03:50] Nathan: That's powerful. Definitely. Definitely. I can't say
[00:03:52] Ty Magnin: that. Um, you know,
[00:03:53] Nathan: I've
[00:03:54] Ty Magnin: perfectly
[00:03:54] Nathan: mastered it, but it is certainly a fun journey to go down.
That is for sure.
[00:03:59] Ty Magnin: Nathan, would you [00:04:00] please give your intro for the audience?
[00:04:02] Nathan: Yeah, absolutely. So, um, I'm the founder of Lex. And we're building basically an alternative to, to like Google docs or word or wherever you write that has AI baked in to really help you throughout the writing process and something that maybe is a little bit unique about our approach is this is coming from the perspective of like, I love writing.
I'm not trying to avoid writing by using AI in my writing process. I'm trying to have a more satisfying writing process. Sure. Maybe a little bit faster in some ways, a lot less agony. Um, but in the end, I want to come up with something that I feel really proud of. personally. And so our belief is like, in the future, AI will play a huge role in like the world's best writers, poets, everyone, maybe not everyone chooses to the same way that some people choose to write longhand and they don't want to write on a computer, but it's like a really valid creative choice to be working with AI when you're writing, not something to be ashamed of or afraid of like a lot of, um, you know, or like look down upon a judge or whatever.
Like a lot of people, I think do now, I think there's tons of. Sort of fear, uncertainty and doubt around AI in [00:05:00] the writing process, you know, kind of for good reason, because there's like a shallow way to look at it. That's like, ah, this is slop. And then there, but there's so many other ways to work with AI.
There's so many different roles that AI can play. And so that's something that I personally find just really, really fun to work on. And there's just kind of the aspect of it. That's like, you know, as someone who used Google docs. As my primary work tool for a while, when I was back, when I was, um, co founder of Every and, and, and president there, I was like writing a weekly column and editing a whole bunch of other people's stuff.
It just feels great to have the chance to solve all the little things that I found annoying about Google Docs. It's cool to have a playground to be like, ah, what if we did something better with like, you know, suggest changes mode or whatever. So that part of it's really fun too. That's nice.
[00:05:39] Ty Magnin: Yeah. I played with the tool not too long ago.
So slick, right? You know, everything I love about Google Docs is still there. Um, you're not trying to like reinvent the wheel in that regard, you know? Yeah, I use it a lot and I
[00:05:50] Tim: would love to get into it. But I, I, I just want to back up one step before we get into all that because I think, I mean, just your career, man.
We could talk about that for a whole episode. So I, I [00:06:00] just want to touch on like, can you maybe tie together a little bit like one or two things that you've learned at like, you know, you've been at Gimlet, you've been at Substack, you've founded Avery, now you're doing Lex. Like what, what are you taking away from those?
earlier roles that you're using today.
[00:06:14] Nathan: I feel like it was just incredibly, incredibly fortunate to be in a position to like, sit in like a gimlet media, to sit in the edit room, to watch Alex Bloomberg, this guy who, you know, came up in this American life, produced amazing episodes there, then started the planet money podcast, which is incredible, incredible storytelling about the economy to then, you know, co founding gimlet, like startup podcast is like one of my favorite, just in general, I would say pieces of media to be able to be there and like, see kind of.
How the it's not sausage. It's like beautiful how it gets made. It's like sausage is doing a disservice, but how the sausage gets made is, um, was really inspiring. And then kind of in a parallel way at Substack to see like, you know, Chris and the product decisions you make in Hamish and the way that he relates to, to writers and [00:07:00] garage, the CTO who's lesser known, but just incredibly important to Substack seeing like, this is what amazing engineering looks like.
I just feel. Lucky to kind of observe genuine, like genuine excellence looks like in sort of a lot of different domains, you know, cause I think it's okay. It could be cool if you're like a really technical person, maybe you get a chance to see excellence in like engineering, but like, can you also be in a meeting with like excellence in storytelling, right.
You know, and just to be able to at least soak up some percentage of that, or also have the courage to be like, yeah, it does look actually quite messy in the early phases. And that's totally fine. It doesn't mean that. We're dumb or this thing doesn't have a chance in hell of working. It means like, that's just how everything starts.
And you just got probably the main lesson is like, it's not really magic. It's just trying a little bit harder, a little bit longer, you know, caring a lot and, and not letting anybody tell you it's not a good idea to care. Being honest, even when something's a little bit hard, uh, all that stuff is just, there's not like really a secret to it.
It's just. Application of that consistent effort over a long period of time [00:08:00] and trying to stay as focused as possible. Um, so seeing how it's not, not really magic in the end is like that. That was really cool. Um, those are probably some of the main, the main things that I took away from those. And then there's all sorts of specific stuff about like the writing process or whatever, that, um, kind of leads into the philosophy of like how Lex is designed and things like that.
But like the general stuff is just, yeah, that's not magic, which is cool. Yeah.
[00:08:23] Ty Magnin: Yeah. Anybody could do it. Just got it. I like it hard work and focus for a long period of time. It's a very American
[00:08:28] Nathan: philosophy. Yeah, yeah, totally right.
[00:08:31] Tim: You know, you were at Every, and then like, how did Lex start? And like, what was your, when was it like, I guess the AI moment where you thought, we need an AI writer.
[00:08:38] Nathan: That's funny, there was never a moment when I was like, we need an AI writer. Like, it's, I feel grateful that um, it didn't come from a place of like, Hey, like AI is happening. Like, what are we going to do or whatever? Cause it was the summer of 2022 when I very first started thinking about Lex and it was like kind of the late, it was like September ish when I actually started building it, uh, which was pre chat GPT [00:09:00] that was like, um, people were talking about image generators at the time and like the cool thing with AI was like.
Oh, cool. Like, you know, we, we could never have really afforded to have cover art illustrated for all of our posts, but we could like get AI to generate it. It was a cool example of like, just AI producing abundance rather than necessarily only just like taking people's jobs or whatever it's like, it was, we never, we never could afford it before and we didn't do it.
So that image generators was what was on kind of people's minds in summer of 2022. But for me, I just, honestly, I missed building software. Most of my career has been, you know, designing program, all that kind of stuff. And there's an, to an extent, some of that at every, but mostly my job was to produce a column every week and to try and make it go viral, which started becoming increasingly difficult for me.
Cause I was like, I feel like I'm running out of things to say and. To edit other people's columns and like recruit writers and things like that. You know, I went from using like whatever VS code now I use cursor, but at the time I was using VS code and like GitHub and Figma and stuff like that to like Google docs.
And I just felt this steep [00:10:00] drop off in sort of quality or thoughtfulness of, of the tool. And like, you could kind of, people did it, like they relied on it, but I just felt like there was a very weak tie to it. You know, there's like the tie of habit, the tie of custom, but not the tie of like deep, like this is a thoughtful tool that's made to do what we're trying to do.
It's kind of a general purpose document editor used by a million different people for a million, actually a billion different people for a billion different reasons, almost it seems. Yeah. Just thought it would be how hard could it be to make our own thing that we can use internally and just gradually make it better over time.
And the primary motivation was like, I missed building software. I was annoyed at Google docs for small reasons, honestly, like. Very comically small reasons. Like, uh, if you like do the keyboard shortcut to bring up the emoji picker, like doesn't work in Google docs for whatever reason.
[00:10:44] Ty Magnin: What's up with that?
And
[00:10:47] Nathan: things are a little bigger. Like the mobile app, if you're in suggest changes mode, you type out key. And then like five minutes later, the letter shows up on the screen, you know, like things like that. And then kind of bigger, more philosophical [00:11:00] things like how do we make exploration and branching and versioning in some sense like easier without making it feel super complicated?
And then like just the question of how AI would play a role. It definitely was on my mind and it was a key part of flex from sort of day one is like the thought of like something with AI could make sense here. But it was more just like, um, uh, curiosity than it was like a specific vision. I wasn't like, we have to, and honestly, I didn't think it would be like, necessarily the most important thing.
I thought it was just like, and, and a fun little thing, you know, more so than like, the reason the company exists or whatever, um, or that it became an independent company. So it was just a side project. Worked on it for like nights and weekends, basically for about a month and then launched the first version of it.
And the feature we chose as like the standout thing to like build the whole launch around was this, this idea called plus, plus, plus, where when you're typing, if you're stuck and you don't know what you want to say next, you can type plus, plus, plus, and the AI will fill in the next paragraph. So it's like autocomplete, but it's, but it's only there when you want it.
And it's a little bit beefier than what autocomplete, which normally, you know, especially at the time would be like [00:12:00] a couple words or whatever. And the idea wasn't really that you would like use that or you would use it to write a whole post or whatever. It's just like, hey, it's something to do other than check Twitter.
If you're feeling stuck when you're writing, which is like a common, a common feeling for me. Um, and that just exceeded all expectations. It was like a crazy 24 hours. We had 25, 000 signups in the first day. And then it just kind of kept going from there. And it was like, whoa, clearly there's something happening here.
And like the, what the something was, it was very clear to me, which is it's not. It's like 90 percent about GPT 3 was much better than people realized and 10 percent like I did a good job sort of packaging it and connecting it to a problem and kind of like connecting the last mile when not a lot of people had done that yet.
And then, but, and so, you know, it kind of died down after that some, cause it, well, it was just sort of side project quality. It wasn't really viable to replace in your workflow from, from Google docs, but just kind of chipping away at it. The hundred thousand mile March thing, because I was just like, I just can't stop thinking about this.
It's the most fun I've ever had in my career. [00:13:00] My whole theory about it would be fun to be working on software was like validated. And then some, it was like so refreshing to be like, yeah, this is the kind of work I want to be doing. And, um, it gets to bring together a lot of my kind of idiosyncratic past experiences to be, to be something that I feel.
Uh, kind of like uniquely suited to build and, and just most importantly, I'm having so much fun doing it. So yeah, that was, it was about, um, two years ago now that we like started the kind of spin out process to turn it into an independent company. Raised a little bit of money after the spinout was done, which ended up taking a while because you know, spinouts are complicated.
And then, uh, yeah, I hired a small team. So it's been a year and a couple of months since it's been more than just me working on it, which is amazing. And, um, yeah, it just feels kind of like we're off to the races. There's a really good, like financially, we're probably going to get to profitability this year, which is great.
And it just feels very like sort of stable and. You know, there's a lot of companies I look up to that have had this path of like not getting too far ahead of themselves. It was like the fundraising and the valuations and all that stuff, but at the same time, raising [00:14:00] money and like going for a big, you know, like trying to, trying to solve a big problem that could reach a lot of people in the end.
Um, you know, like Zapier comes to mind, uh, GitHub comes to mind, that kind of thing. So, um, Linear is maybe another one. Notion is probably another one, so. I
[00:14:16] Ty Magnin: feel like there should be a name for that kind of startup. It's not Bootstrap. It's not VC backed. It's like in between, you know?
[00:14:21] Nathan: Well, it is VC backed.
It's just not like getting too far over your skis, which is so tempting to do. We're like, the VC is willing to value you at 500 million. Even if your revenue is only like. It's kind of like tempting to just be like, yeah, let's get this, all this cash for like, so little dilution. Um, but yeah, it's like,
[00:14:39] Tim: not, not the rocket emoji kind of thing.
At least that's what it's not. Cause of course I would recommend everybody to read your annual letter. I think it's a great read for any content marketer and writer interested about AI. So, so we'll link to that in the show notes. I think one thing I liked when you touched on a little bit that's in the letter, it's like you, you talk about the comparison between writers and programmers [00:15:00] and the differences you see, cause you've, you're obviously.
Close to both. Can you talk more about that?
[00:15:05] Nathan: Yeah, so the title of the letter is, Why Don't More Writers Use AI? And for us, the reason why it's kind of like our annual letter is because when I was stepping back to think about like, well, how far have we come? And like, what's still in front of us? The answer was like, we're at the very, very bottom of like the S curve of adoption with writers and AI.
You know, like most writers, maybe they use Grammarly, maybe they pop over some questions or chat GPT like every once in a while, but it's not like they've really adopted it as a part of their writing workflow in any real deep way. And so I just wanted, it's like important to think about like why, you know, like if that's our goal is to, um, make AI more useful.
And then especially to have, cause it's always hard to evaluate these things in just sort of like no context land, but to kind of anchor it against a comparison point. So programmers. Like, in my, in my job, when I'm writing code, I'm using AI all the time, every day. [00:16:00] Cursor just raised like a giant round and it's getting adopted massively.
Google has reported like 25 percent of the code that their engineers check in who are like, arguably, like it's probably a pretty talented group of engineers, pretty experienced group of engineers. They're checking in 25 percent of their code is generated by AI. Like nothing like that level of adoption is having in writing.
We're just not there yet. So for us, like. Sort of from a selfish personal point of view. It's kind of nice because even though we haven't achieved our vision like nobody else has So it feels like it's open, but at the same time like, you know What really matters is just sort of like why and how do you actually make AI more useful to writers?
And I kind of grouped the reasons into two categories One is like just conceptual hangups of like, until you kind of have the, have a certain perspective on AI, then you might not even want to try. You might be like, this is, I should stay away from this. And then the other, the other is just practical difficulties.
I think there's a lot of ways in which AI is much more suited to automating programming than it is to automating writing. And so, so starting with the kind of [00:17:00] conceptual. Hangups, one of them is just sort of like, is it, is it weak? Is it a weak move to use AI? Should I admit if I'm using AI, am I going to get in trouble?
Am I just going to be judged? Should I be embarrassed of using AI? And you can kind of break it down into like, sort of, is it sort of, is it ethical or, or, or, or authentic or inauthentic or unethical? And then the other one is kind of like, is it just a weak move? It's going to weaken my writing ability. So the way that I've thought about it is.
Using AI is like hiring someone, but there's a lot of different roles that you can hire someone to play. So, if every time you needed to like lift anything around your house or do anything, you hired someone to do that for you, like yeah, your muscles would atrophy. Um, but at the same time, you can also hire someone to be like a personal trainer, right?
And make you a lot stronger. And AI is kind of like the same way. You could just be like, anytime you have to write any little thing, like AI, write it for me. You can also say, hey, here's my writing, like critique it. You know, there's a whole bunch of different roles you can ask AI to play. So that's sort of like, can it make you weaker?
Yes, but it can also make you stronger. So the, the authenticity one is like, [00:18:00] you know, is there sort of wrong or weird to like use AI in the writing process? Like it feels kind of obviously bad for a student. Like there's some cases where it just seems obvious, like no one says it's a good idea for a student to like generate an essay and turn it in and say that they did it.
Like there are whole companies built around this and they use all sorts of sort of like double speak to be like, it's for academics, you know, but really it's for students to cheat and like they're growing super fast, you know, and so it's like, uh, there's obviously demands for that. Or like, you know, just sort of like, ah, generate like spam, cold sales, email type stuff on the fly.
And it's like, that's fine too. And there's nothing inherently wrong with cold sales email, but like. You know, there is something a little bit different about that from, uh, using AI to like, whatever, right. Something that like matters to you. It just all comes down to how you use it. And there's many different ways to use it.
And so you can say like, Hey, I've got this template and then now here's all this stuff about this person that I'm trying to reach out to. And can you give me some ideas of like, what might be interesting to them here? And then you like, think about it and you write it and like, that's great. And there's.
[00:19:00] Also ways to not do it that well. So anyway, I think it's, it's all in how you use it.
[00:19:03] Tim: Well, and, and I would, I should just say, because I'm familiar with Lex, but for people who are not familiar with Lex, one really nice feature, I think it's very simple, but it's really cool, is that just like in a Google Doc, you can leave a comment, you can actually leave a comment for Lex.
So you, you tag Lex, the AI system or the, the tool in the, in the comment, and then it will actually answer whatever you, you say in that comment. So that's a really thoughtful way of like, you know, you select a paragraph or whatever. And then indeed the kind of question that Nathan just mentioned. Um, you can pop those out, like critique this or what, what am I missing here or whatever.
So, so that, yeah, that's, that's really nice. I think how you, how you integrated that into the UX of the tool.
[00:19:37] Ty Magnin: So what I'm hearing is there's kind of like a spectrum of ethics in a way of like, Hey, like you don't want your. Teenage son to be using this is a real use case, by the way, I've got one of those to be using a I to like, you know, write your paper for you, right?
Or like, because you want your students potentially, I don't know exactly about where education is going. And I haven't thought through this whole [00:20:00] thing, but like, I don't know, I kind of object to that idea, right? Versus like, if you're using it in a way to help that person do research or like write a better paper, and then it's like, all right, it's just kind of like how you use it more or less what you said.
I also just wanted to relate. To this fear that you've mentioned that writers might feel of how others might perceive AI usage.
[00:20:20] Nathan: Yeah.
[00:20:21] Ty Magnin: Okay? Because I'm always a little sensitive, uh, when talking to customers and animals of like, Hey, like, here's how we're thinking about AI. Here's how we might be using it today.
Like, and I'm kind of like. Does that jive with you? Cause some are like, no, we want our agency to be figuring this out for us. Like, that's a good answer for us. When the C level is coming, knocking on our door saying, use more AI. Sometimes you do run into someone that's like, Oh no, we don't want any of that.
Like, we don't think it's good. And you're kind of like, all right, there's like an education opportunity here. But my point at large is, um. Because especially writing type [00:21:00] folk or content folk are at different points in their adoption journey and their kind of emotional relationship to AI usage. It is a bit of a hot button issue as content marketers to kind of, you know, say we're using it.
I think it's becoming less so as more folks are finding their way around that adoption curve. But that's like a different audience than developers I'm going to argue too. Developers are technical people and you talked about this in your essay. They tend to be early adopters, right? So therefore, like, it doesn't surprise me in a way that they might be further along the adoption curve as well.
[00:21:35] Nathan: Totally. Programming is all about remembering obscure things, and people, when you're programming, you're constantly looking up in documentation or whatever, it's like, what's the exact words I have to use here? And you kind of know the general idea, but like, when you're writing for people, Like that, you're just writing for other people's brains.
It's sort of like, it's pretty native. You don't have to look up, like, what's the word for this thing exactly again. You do that sometimes, but it's not nearly
[00:21:58] Ty Magnin: as. There's a thesaurus [00:22:00] for a reason. Yeah,
[00:22:00] Nathan: exactly. Like you do it sometimes. And like, we have a thing in Lex to do it, but it's less like mission critical.
Like if you get the slightly wrong word, they only get a slightly wrong impression versus the code doesn't run. And when you have like a slightly wrong. Yeah, that's black and white. Yeah. Yeah. So I think like remembering these really obscure incantations and like formulas or whatever is like, just AI is great at that, you know, and so there's a lot more and like, you know, with, with programming, like you can, you can just see if it works yourself by running the code, you know, whereas to test, you know, stuff that AI wrote, it's kind of like, well, how do I feel about this?
How would someone else feel about this? Like, it's much more, you have to read it, you have to think about. If it works or not. So it's, it's a lot trickier to kind of verify or validate the, the solution that is coming up with, which is why we honestly, I think it's great for AI to prompt people and a general principle of ours is like help you think about what you're doing and think through and give you stuff to potentially do.
of ways to improve it, suggestions. Like when a good editor is like, I'm thinking this. And you're like, they didn't tell me exactly what to write. But you're like, oh yeah, I need to run with that. You know, [00:23:00] like that's a great role for AI to play. Much easier to do that than like, generate the final paragraph for me or whatever.
You know, and like, it can try, but yeah.
[00:23:08] Tim: I think another issue is that any programmer that I know is always obsessed with improving their productivity and almost every content marketer I know doesn't care at all about, or most of them are always like, they just do it always in the same way. So I think there's also this thing of like, you know, programmers are engineers, right?
And, and then, and then we have an article somewhere in the blog. It's kind of the essence is kind of like, no, you're not a writer for like content marketer. You're a content marketer. Right. But I think a lot of content marketers are, see themselves much more as creative and whatever. I think AI comes also more natural to an engineer because it becomes a little bit more about building a workflow.
Whereas a lot of writers don't see it as a workflow. They just see it as like. Being being creative. So there's also that kind of kind of mindset. Is that something you see?
[00:23:50] Nathan: I definitely agree I mean there's kind of a masochism in writing of like the harder it was maybe therefore the better the output was that I don't Think is like comes from when you are very particularly healthy, you know [00:24:00] And like, you know, sometimes it's there's something really to it Like yeah It's great to do more revision or whatever and like that usually helps if you put a little extra thought into it and all that stuff kind of like what I was saying earlier like what I what I learned at Gimlet also Another thing I for sure saw at Gimlet was people agonizing over stuff that like, maybe didn't matter that much.
A lot of anxiety, a lot of wheel spinning. You know, you read, you read about like, you know, people with like writer's block or whatever, it's like, there's a lot of stuff in the culture of creative endeavors that, um, I don't really think serves anyone. And so I think there's something very true to that. Um, just, just culturally.
The other thing I will say though, is I do think you can use AI to do all those things, not that I would necessarily endorse it, but you can have AI. Spit out a million different critiques of your thing. Right. And then you can go agonize over and create a million different versions and whatever you can do all the creative wheel spinning with AI, just the same way you can do without it.
You know, not that I necessarily think it's like the best thing to do, but yeah,
[00:24:56] Tim: that, that's nice. Cause I think that ties into something I would love to hear from you. It's [00:25:00] like people very quickly connect AI to like efficiency gains, but like you have good examples of where it actually improves quality because some people don't say that, but there's not.
That many examples of it? Yeah. Where does it significantly improve the quality of your writing?
[00:25:14] Nathan: Yeah, it's hard to have a counterfactual because like usually you don't produce like the non AI version and then the AI version or whatever, you know what I mean? Like the AI enhanced, it's hard to know what you would have come up with in a world where AI didn't exist, you weren't using AI or whatever.
Um, I can look at all my Writing from before AI, and then look at my writing now. And like, for instance, this piece that I just published the annual letter, like, I think it's a lot better than a lot of stuff that I wrote before AI. Also, is it just cause I'm a more experienced writer now? Is it cause I had some time off and I like felt more spaciousness to do this.
It's hard to AB test. To me, the thing I look for is there's a threshold of experimenting with it and trying it that once people kind of like use it to a certain degree or feel the value. They don't tend to go back. And so it's kind of like, you can just tell if [00:26:00] someone thinks it's helping them, like they keep doing it and.
You might try it once and not feel like it helped that much. We're like, but once you get to that level of using it, where you feel like it actually helps you, like it. You just tend to do it, and maybe you don't do it in Lex, right? Maybe you're like, ah, I just want to use ChatGPT or whatever. It's the retention, I think that is probably the main empirical way to measure it, which is not exactly measuring if it improves the quality, but it's, does it, does the writer think it improves the quality?
Uh, to some extent, probably. So, yeah.
[00:26:27] Tim: I like your idea of comparing it to something you wrote before there was AI. That's a good one. What about you, Ty? I think
[00:26:33] Ty Magnin: you have a question. I mean, I have like the dumb obvious question, Nathan, of like, how are you using Lex in your workflow?
[00:26:39] Nathan: Totally. Well, the first thing I do, which I didn't do as much, honestly, before AI, Is I do a brain dump and, and the reason why I find that useful, even honestly, AI aside is just because it helps unpack my thoughts a little bit.
And it gets me out of like, I'm writing a draft and then to like, I'm writing about what I'm going to write about kind of is not an outline. It's just like, what are my goals? [00:27:00] What are the main things I want to hit on? Where did this come from? It's kind of like free, right? A brief journal. Yeah. Okay. Okay.
Not probably as structured as a brief, but like a brief basically, like a brief, but like kind of just for me and allowed to be really messy. If it was a real brief, I would probably, you know, have to edit there or something like that. But yeah, and that is great for AI, because now you can reference that and your draft and be like, Hey, is there like, am I hitting the mark here?
You know, it gives, it gives the AI that context. That if it just has the draft, it can make assumptions, but it's much better if you give it the sort of, and then what I also do is I'll copy paste tons of stuff. So if there's like quotes or references or like a paper or something or an interview transcript.
Or like a whole bunch of customer feedback or whatever, just throw all that at the AI and then, and give it your draft and then ask specific questions. So that helps a lot. Um, that's kind of like the first step of the workflow is the brain dump, the brief. I like that. And then the, the next step is like, I'll write a first draft and I, I almost never outline personally.
Because I [00:28:00] like to have a first draft kind of like just come out. And then I go back and I work reverse outline and sometimes it's easy to do with AI, like, Hey, here's my draft. Can you just give me what the outline of this basically is? You know, um, it's one of those tasks that's like a little bit menial, but also maybe it comes up with something I wouldn't have, and then maybe I'll come up with my own outline.
We actually have this new canvas feature that's coming out soon that I, I think Tim has seen that's, uh, basically you can have all your versions laid out like on almost like it's Figma where you can like zoom in and zoom out and rearrange pages and you can put little labels on the canvas that aren't like, you know, a part of the document it's like just next to the document.
So I actually started labeling. And like the margins almost as if I'm scribbling, like on, on a whiteboard or paper or something like that. So that's a cool way to do the reverse outline thing. And I can kind of zoom out, like literally, cause it's like a design tool, almost resuming out. Um, and look at like, okay, what was the flow of thoughts here?
And like, could I kind of get, is there a better flow of thoughts or could I achieve that flow in like a more efficient path? And maybe I'll also share it with people, like a teammate, um, I'll, I'll ask, maybe there's some specific things that I'm a [00:29:00] little worried about, or I'm not sure if this part's great, I'll ask AI about it, ask a teammate about it, and then usually I have another document, that's like my notes and thoughts, and like to dos, or whatever, and I'll just kind of consolidate all that, and then what I'll do is I'll just like take it from the top, I'm not even gonna like edit my previous thing, it's like a whole new page, blank sheet of paper, let me just start writing this thing again, and now I've got more of the like, I've got the outline, Roughly in mind.
So I can usually flow through it really quick and I'm way more efficient about it because I can just pan over. On the canvas copy and paste individual sentences or whatever, or paragraphs where I'm like, ah, this cleanly slots in here, but I'm kind of reassembling it from scratch in a way. So there's no like loss aversion of like, ah, do I cut this or whatever?
And that's kind of almost like the first real draft in a way. And then from there, I just do a lot of specific, like, you know, wake up. Usually I try and write first thing, first thing I do fresh, fresh brain. And then it's kind of like, all right, like take it from the top. For the most part, you know, if I was writing something longer, I probably couldn't take it fully from the top.
It would be like to take from the top of this [00:30:00] chapter or something like this. But, um, some specific features of Lex that I like to use, um, in the phase where I'm actually drafting are like, like, for instance, um, the, the comment thing that Tim talked about, where you just, you select some text. You hit, uh, the comment button at, you hit ask Lex, and then I'm just like, anything from, are there other historical examples of this?
What are some other analogies I could use or metaphors? Um, is this right? Uh, would people disagree? Would a lot of people disagree with this? Like where, or like at the overall chat, which is like not about one specific part. I ask things like, yeah, like what are, what are the things that people, where people might get confused here?
What are things that people might get bored? Are there areas where people might not agree? There's this acronym that I love that's like ABCD when you're looking for feedback from, from people and AI is great at it too. What was awesome? What was boring? What was confusing? And what didn't you believe? And didn't believe works really well for fiction and for nonfiction.
Just like disagree or like, I don't buy it kind of a thing. And so AI, AI is really great at that. [00:31:00] And I'll just go through a bunch of iterations where I'm just maybe one in five things I'll actually think is valid, but I'm kind of, it's more like mining than I am testing. Like I don't need the AI to bat a thousand for me to find value out of it.
If I'm asking it for like a hundred suggestions in the course of drafting something. And like, 30 are great. Like, great. That's obviously, I'm gonna improve. And some of them are like, it's a one liner, but I'm like, oh, yeah. And that like, has big implications for the draft, you know.
[00:31:26] Tim: I can say Lex can be, can be pretty strict.
Like, I feel Claudius always wants to please me. And Lex, like, I'm like, I'm always like, oh no, here comes Lex feedback. And so it's like, it's not good enough. I'm like, oh no.
[00:31:37] Ty Magnin: It's a tough editor, that Lex.
[00:31:39] Tim: Yeah, really, it's a tough editor. Um, well, most days it's a good thing. Sometimes I can't handle it. It's like, okay, I'm going to, I'm going to leave that.
[00:31:47] Nathan: We edited our system prompt to say, um, because of your training, you will be inclined to be really nice to people, but you'll actually in so doing be doing the writer a disservice. They want you to hold them to a high standard. You know that [00:32:00] they can do better than this. So help them see that. And it's like, it really changes.
The AI is like approaching it. Yeah, it helps it come up with good stuff. Whereas like, you know, stock clot or stock trajectory is kind of like awesome, you know, you did great. Here's a couple things I liked about it or whatever. And it's kind of like, you
[00:32:15] Ty Magnin: have to tell Lex to take the gloves off. It's
[00:32:18] Nathan: like, I do
[00:32:19] Tim: the opposite with clothes, whereas like, be honest, please be honest.
Or I also did like, I'm asking for a friend. Don't worry about me. Like, just tell me the truth when you do the opposite in the background with Lex. That's pretty cool. I quickly before because we're running out of time, but I would like to zoom out like. One more time, a little bit and, and, and get your thoughts on how do you think about reading?
And I ask that because, you know, ultimately we're writing so that people will read it. And I think there are many voices, I think, you know, in content marketing, other places who might even say that like reading is kind of. Going out of fashion, right? Whether it's because video is becoming more popular or because AI will just summarize whatever we write.
How do you think about that?
[00:32:58] Nathan: I don't know. I just published, um, [00:33:00] like a pretty long essay to our email list. And like, it was, it made a big difference. Like in our numbers. Immediately. If that's not proof that at least as of today, reading still matters, then like, you know. It, and especially reading when you like, really, I think what doesn't matter is things that it's just like, people are going to ask an LLM.
Like this is why I always thought that LLM is just generating content that someone copies and paste onto a webpage or whatever is so self defeating because the readers will figure out I didn't need the middleman to like ask a prompt for me and copy paste the output. I'll ask the LLM directly and like, you know, the AI tone of voice, like doesn't really matter there.
Cause it's just, I just want the information or whatever. I mean, to kind of zoom out, like what are people really trying to do with thought? Leadership content or whatever, like they're trying to lead. They're trying to say like, here's a perspective. And so like, you know, the opportunity for us, I think is, um, writers are curious about some writers are curious about AI, um, but maybe they've also got some fear, uncertainty, doubt.
And like, they, they, they're want to learn, like, could I actually use this? How could I use this? Is it [00:34:00] okay to use this? And so it's an opportunity for us to say like, well, here's what we think. You know, if this helps clear it up for you, like, great, we've, we've got a tool you can use. We're obviously going to write from the perspective of like the tool we're building and no matter what kind of business you're building, there's going to be some stuff that your audience is uncertain about, is excited about, but is maybe a little bit scared about.
And that's what leading is about is helping people through uncertainty and confusion and, and sort of navigating between like hope and fear, basically, or, or greed and, and fear. Like AI itself is not going to be able to really lead. It can kind of regurgitate a little bit, but it doesn't, it can't like lead.
And so that's for people to do.
[00:34:38] Ty Magnin: What a beautiful place to end. You know, we're kind of on the same mission. Uh, maybe we're not writing so much with this podcast season, but we are focused on helping folks. Learn their way with us through AI and the changes and figure out where the value is and how to use the tools So in some sense, we're helping support that mission, but we're listening today.
We're talking through these things [00:35:00] Rather than writing. Amazing. Where should people follow you?
[00:35:03] Nathan: I'm on I'm on Twitter slash X Whatever we're calling it these days NBA SH AW Um, and then at lex. page is the company handle and lex. page is our URL. So if you want to try it out, uh, would love to hear any feedback you have, any questions, it's hello at lex.
page is our email and feel free to reach out to us directly if there's anything we can help with.
[00:35:25] Tim: When are you launching a blog? That's my final question. I was looking for your blog and I can't find one.
[00:35:29] Nathan: Yeah, we don't, um, we're going to have to start doing that. Cause honestly, this, uh, was really like re energizing for me to like put.
You know, substantial effort into a piece of writing and publishing it. So we're going to start doing it more often. And, um, well, if you, if you sign up for Lex, then you get added to our newsletter, which you can unsubscribe from, but you get to it. And so, uh, just sign up for Lex and then you'll get emails about, uh, what we're thinking and, uh, we'll probably announce some sort of blog ish type destination soon.
[00:35:57] Tim: Awesome. All right. Thanks so
[00:35:59] Ty Magnin: much,
[00:35:59] Tim: Nathan. [00:36:00]
[00:36:00] Ty Magnin: What do you think? That was great. That was great. You liked it? Yeah, I liked so much of what he had to say. One, like, an honest shout out, like, Lex is a great tool. I'm a, I'm a superhuman user. I'm not using Lex right now. But it reminds me of superhuman if anyone uses that email tool along with me.
It's just like, really well designed, simple to use. You don't have to learn it. Like, there's a, the learning curve is short, you know, and then you kind of get addicted to it. Like, I think Lex is in that category. And yeah, Nathan, I mean, man, what a resume had so many great ideas to share and look forward to learning from him, you know, as they start this blog and he continues to kind of lead through this big AI change.
[00:36:40] Tim: Yeah, no, I mean, he's the kind of person like I wish we had a three hour podcast, right? Because he's just talking about his career. It's like that Gimlet media, what he mentioned, like, I listened to startup. I love that also, like, as he described it, it's like, wow, you know. Indeed, like being in there with those people and building that, like just that.
But of course, that's not the topic of what we're talking about. The difference between programming and [00:37:00] writing. Things there where I've also been comparing writing to programming and how it's different. And so I think that's, that's super interesting. You know, he's just ambassador for writing in a way, which is also cool.
Right. He's just really excited about writing, which I think is also
[00:37:13] Ty Magnin: refreshing. Yeah, it's
[00:37:14] Tim: nice.
[00:37:15] Ty Magnin: I am also a believer in the zoom out method of writing content or editing content. Okay. I like to zoom out on my Word doc or my Google doc or whatever I'm drafting in, in order to kind of see like the shape of a piece.
There's something about the rhythm of content that you can spot by looking at the aesthetic of it. Is this paragraph too fat, right? Is this section, you know, too long? Is the headline too long? Like, how does it look when I zoom out? And so I like that Lex has that feature built in. Uh, again, it's like a really nicely designed.
Writing interface, but zooming out on that,
[00:37:50] Tim: I think what he does very smartly, they look at these other kind of tools from other disciplines and say, how can we use that? 'cause they now also have a thing where it's a bit more like a code [00:38:00] editor in the side. When it suggests changes, you actually also see the old version, a new version, and like the new version is.
Kind of highlighted green and what they're gonna remove is highlighted red. That's exactly what Cursor, for example, does. Yeah. They're looking across, you know, at other tools and taking inspiration, working that into the ux. So yeah, it's, it's also really cool. Yeah. Also some frameworks. I like his framework about hiring it for a specific job and how he unpacks that.
'cause I. I think it's kind of a cliche that I've seen thrown around, but he actually has a real practical idea around it. And, and also personalizing that, like where, for where you are as a writer or as a content marketer, does it make sense to use AI? So he makes this distinction of like, if you're already very established and you have your own style, it makes less sense or it's much harder to get AI to write in your style.
So that's then not the right use case. I think, I
[00:38:46] Ty Magnin: think that's a good way to think about it. That does it for another episode of the Animals Podcast. Thank you for listening. We'll see you on the next episode.
[00:38:53] Tim: Yeah.
[00:38:54] Ty Magnin: Bye bye.
