# Transcript — $30M Writer: Never write AI slop again (Greg Isenberg with Nicolas Cole)

- Video: https://youtu.be/YuOSyRj3sXg
- Source: YouTube English auto-captions, fetched 2026-10-10 with yt-dlp, rolling duplicates removed. Not checked against the audio. Speakers are not named; ">>" marks a change of speaker where the captions show one. Names are misheard in places ("ghostriter").
- Stored for review miadi-review:1db9f108-5c01-4c32-ad9a-566371dc9c70 by the Mia seat on gaia.

[00:00] Nicholas Cole's made over $30 million
[00:02] writing on the internet. So when he
[00:04] talks about content, I listen. And this
[00:07] episode honestly felt less like an
[00:09] interview and more like just a free
[00:11] master class. He breaks down why most
[00:14] people's content gets ignored in the AI
[00:16] era, why AI is making generic writing
[00:19] worthless, and how to build what he
[00:21] calls a digital brain, a library of your
[00:24] stories and your opinions and your
[00:26] lessons and ideas that gets you
[00:28] customers 24/7 that builds you trust and
[00:31] gives AI something real to work with. If
[00:34] you write online, you sell online, or
[00:36] you want ideas to compound, this is
[00:39] honestly worth thousands of dollars, but
[00:41] it's free for you. the people that stick
[00:43] around to the end of this uh you know
[00:45] are are not going to be writing AI slop.
[00:48] And I think the people that you know use
[00:50] AI to actually create interesting
[00:53] content I mean of course those are the
[00:55] people that have the highest probability
[00:56] of success. Enjoy the episode. I learned
[00:59] a lot. I think you will too. Today's
[01:01] episode is brought to you by Brex. My
[01:03] company's been on Brex for a year and a
[01:05] half and I started because I kept
[01:06] hearing companies like Verscell, OpenAI,
[01:09] Anthropic were using Brex and I figured
[01:11] if they're using it, why shouldn't I?
[01:13] It's been a gamecher. The thing that got
[01:16] me is how smooth it is. It's got high
[01:18] limit cards. It's got banking. It's got
[01:20] AI that handles the back office busy
[01:22] work like expense reports, which I don't
[01:24] want to do on its own. It's really just
[01:27] built for this agentic world. If you're
[01:29] building something new, it's time to get
[01:31] B. Check it out at brex.com/solutions.
[01:35] Link in the description.
[01:40] [music]
[01:45] >> Nicholas Cole, thank God you're here,
[01:47] man. By the end of this [clears throat]
[01:49] episode, what are people going to learn?
[01:52] By the end of this episode, you will
[01:54] better understand the short, medium, and
[01:57] long-term value of everything you create
[02:00] and write and market on the internet
[02:02] before you create it. That is the goal.
[02:05] Before you invest the time, you will
[02:06] understand why you're doing what you're
[02:08] doing in the first place.
[02:10] >> Yeah. And I think that matters a lot
[02:11] because all of us are, you know, vibe
[02:14] coding software. We know there's a lot
[02:15] of opportunity to build apps, ship SAS,
[02:19] things like that. But it's hard to get
[02:21] customers. And what I love about you is
[02:24] through writing, you understand how to
[02:27] get seen, how to create stuff that goes
[02:29] viral. And there's sort of a bit of a
[02:32] art to it. And what I hope to learn to
[02:35] you know selfishly by the end of this
[02:37] episode is to to to get some of that uh
[02:41] art from you to learn about that so that
[02:44] you know the products that I create I
[02:46] could accompany it it with content
[02:48] writing that's tells stories that you
[02:51] know hopefully goes viral or that just
[02:54] resonates with people and um Cole can
[02:58] you commit can could you commit that to
[03:00] me? Can will I learn that?
[03:01] >> Yeah, that's what we're going to talk
[03:02] about. I I so some something I see
[03:05] people talk about a lot these days is
[03:07] that in the AI era SAS no longer has a
[03:12] moat
[03:13] and I actually think that that's not
[03:15] true. I think that SAS has a moat when
[03:19] you have a point of view. And I didn't
[03:23] that was a nice little rhyme there. But
[03:24] what I mean by that is if you're just
[03:26] building commodity SAS, meaning like
[03:28] you're just providing features that
[03:32] accomplish tasks to people, um that's
[03:34] fundamentally a commodity.
[03:37] SAS with a point of view means that
[03:39] you're building some sort of product or
[03:41] platform that enables users to do things
[03:46] in a certain way like they're they're
[03:49] executing a certain method. So for
[03:52] example like we have a SAS platform
[03:54] called Typeshare
[03:56] and the goal of Typeshare is not well
[03:59] there's a gazillion scheduling apps and
[04:01] there's a gazillion writing apps and our
[04:03] writing app does writing and scheduling
[04:04] too, right? like that is a commodity.
[04:07] The goal of Typeshare is to quote
[04:11] unquote automate or or make tasks easier
[04:15] for people who want to execute a
[04:18] different methodology for writing on the
[04:20] internet. So there's a point of view
[04:22] that drives the product. And if someone
[04:25] says, well, why would that create a
[04:27] moat? For example, if we say inside
[04:29] Typeshare, well, your goal is we're
[04:31] going to help you build your personal
[04:34] language model, and our point of view is
[04:37] that there's certain types of content,
[04:38] commodity content, personality content,
[04:40] original content. Each of these
[04:41] different types of content creates
[04:43] different leverage inside of your
[04:44] personal language model. Even as I'm
[04:47] articulating that, imagine you're trying
[04:48] to build a SAS platform that does
[04:50] everything our SAS platform does. Every
[04:53] time you build a feature, it's like you
[04:55] have to turn back and look at me and go,
[04:57] "Oh, so but wait, like what does that
[04:59] actually mean? What's the point of a
[05:01] personal language model? What's the what
[05:02] does commodity content mean?" So, you
[05:05] will forever be one step, two steps,
[05:07] three steps, five steps, 10 steps behind
[05:09] because you constantly have to look at
[05:10] the person who's creating the point of
[05:13] view or creating the methodology to go,
[05:15] "Wait, wait, wait. They're the one doing
[05:17] the thinking. I'm just going to copy
[05:19] them." Which means you're always behind.
[05:21] And so I think it's funny when people
[05:22] say there's no moat in SAS because it's
[05:24] like well yeah there's no moat in
[05:26] anything if you don't have a point of
[05:27] view but there's a ton of moat in SAS
[05:30] and product and writing and marketing if
[05:32] you do have a point of view. And so
[05:34] that's kind of what inspired this convo
[05:36] is like so how do you create a point of
[05:39] view in the first place?
[05:40] >> So where do we start?
[05:42] So, uh, an idea I've been thinking about
[05:44] a lot is the fact that the the English
[05:49] language or any alphabet, any language
[05:52] is really just uh an open-source
[05:57] language for communication.
[05:59] Meaning, no one owns the individual
[06:02] letters of the alphabet. No one owns
[06:05] words like the vast majority of words,
[06:07] right? Like you don't own the word
[06:10] tomato. Uh, I don't own the word
[06:12] catastrophe. Um, we also don't own even
[06:16] our names. There are a lot of people in
[06:19] the world named Greg. There are a lot of
[06:20] people in the world named Cole. And if
[06:23] someone says, "My name is Cole," or "My
[06:26] name is Greg," and you write, "My name
[06:28] is Greg," right? You're not plagiarizing
[06:30] them and they're not plagiarizing you.
[06:32] And as you start to like tease that idea
[06:35] out, you realize that the vast majority
[06:39] of words and groupings of words and
[06:42] thoughts and ideas and sentences are
[06:45] shared. They are not actually
[06:47] attributable to any one individual. And
[06:50] I think this is what people are
[06:52] wrestling with.
[06:54] When they read uh writing that they
[06:56] think was written by AI is when when you
[06:59] identify quote unquote AI slob, what
[07:02] you're really identifying is commodity
[07:04] content. Meaning you're reading
[07:06] something but it feels the way that it
[07:09] does because it's not attributable to
[07:10] any one individual. It's like reading
[07:12] health and wellness content. Someone
[07:14] writes uh the key to losing weight is
[07:17] eating healthy. That's a commodity
[07:19] sentence. There's a gazillion people who
[07:21] have said that exact grouping of words.
[07:24] And we don't look at the millions and
[07:26] millions of people who have repeated
[07:28] that same grouping of words and go, "Oh,
[07:29] everyone's plagiarizing each other." We
[07:32] just go, "That's a commodity sentence."
[07:34] And so, as you like start to tease this
[07:37] out, it raises a really interesting
[07:39] question in the age of AI. And the age
[07:41] of AI is the creation of infinite
[07:43] textbased content. Anyone can write
[07:46] anything on any topic at any time with
[07:49] any amount of volume and speed they
[07:50] want.
[07:52] Which raises the question, so what is
[07:54] worth writing in the first place? And I
[07:57] think it's it's interesting even like
[07:58] you your content is a really good
[08:00] example of this where the simplest
[08:03] delineation is when you're writing and
[08:06] you say the difference between I think I
[08:10] did this I perspective versus you should
[08:15] you should do this because when you're
[08:18] talking to the reader through the lens
[08:20] of you
[08:22] inherently the content is going to lean
[08:24] more commodity because you're kind of
[08:27] just like repeating all the things that
[08:28] everybody knows everyone should do. But
[08:31] when you're writing from a place of I
[08:33] like I just did this or I think this or
[08:36] you you share all the time. It's like I
[08:38] think this would be a great startup
[08:39] idea.
[08:41] When someone consumes that content, that
[08:43] is personality content. That that is I
[08:46] am consuming what Greg thinks about the
[08:49] world, not what everyone in this topic
[08:53] area typically says about this topic in
[08:56] the world. You know what I mean?
[08:58] >> Yeah. I mean, that's why all content
[09:00] isn't created equal because sometimes
[09:02] I'll see people go insanely viral on X,
[09:05] let's say, and I look at the content and
[09:08] I'm like, I've read this before. it's a
[09:10] Wikipedia article essentially summarized
[09:12] or it's a just a take that's been said
[09:15] before and they plopped it into chat GBT
[09:18] and asked to create an essay on it
[09:20] >> and
[09:22] uh it it could still get 5,000 likes or
[09:24] 10,000 likes but at the end of the day
[09:27] it's like what game you trying to play
[09:28] when I'm creating content I I think
[09:31] about two things. I think about how do I
[09:34] how do I drive as much value for the
[09:38] person consuming the content. So I'm
[09:40] just constantly as I'm writing I'm like
[09:43] thinking about the person in the other
[09:45] seat and thinking about how can I help
[09:48] them? And that frame of mind is super
[09:51] helpful because that means actually
[09:53] you're not going to write things like
[09:55] you should do XYZ, you should do XYZ
[09:58] because that's kind of coming from a
[09:59] place of like I'm better than you and I
[10:03] know everything therefore you should do
[10:05] this. And I don't think that's what I
[10:06] try that's what I I want to avoid.
[10:09] >> And then the second thing is creating
[10:11] affinity.
[10:13] So uh you know if I add value to
[10:17] people's lives it's sort of an effective
[10:19] number one it it generates some affinity
[10:22] to hey this this guy helped me out a
[10:25] little bit you know I got an idea from
[10:26] him or he you know I never I never
[10:29] thought in you know of uh this topic in
[10:32] that way like for example what you were
[10:34] talking about with like the English
[10:35] language is open sourced. I never you
[10:38] know I use English every day in my life.
[10:40] I never thought about that concept
[10:43] before. And for the rest of time, I'm
[10:45] gonna be associating that with you,
[10:47] Cole. And I think like people undervalue
[10:52] how valuable it is to have a new idea
[10:57] and ship that value to someone and it
[11:01] actually affects their day-to-day,
[11:03] especially within the context of a
[11:04] business context because that means that
[11:06] they're going to be making more money.
[11:07] they're going to be being more
[11:09] productive and that's going to put food
[11:11] food on the on their plates, feed their
[11:13] family, get them a job promotion. Like
[11:16] there's so much good to that and that's
[11:18] what I want to unpack today is like that
[11:21] piece for people.
[11:23] >> Yeah. So, okay, a ton of thoughts.
[11:26] First, VA value is always relative to
[11:31] where the the audience is. meaning
[11:35] something that you could find really
[11:36] valuable because maybe you're a more
[11:38] advanced entrepreneur. Um, a beginner
[11:41] would not find as valuable because
[11:42] they're not far enough along yet. And
[11:45] uh, in reverse, something that's very
[11:47] basic to you that you've known and
[11:50] you've learned that lesson a thousand
[11:51] times um, is not valuable to you anymore
[11:53] because you already know it. But to a
[11:55] beginner, that's like the most helpful
[11:57] thing you could give them is step one.
[12:00] So value was always relative to where
[12:02] the other person is. What prompted this
[12:05] whole mode of thinking for me is
[12:09] especially in the past like 2 or 3 years
[12:10] and as I've seen AI start to accelerate,
[12:14] I've found myself living in a constant
[12:16] state of frustration where I know that
[12:18] these models can write a lot of things
[12:21] just as well as I can. My frustration is
[12:25] knowing that that that uh solution
[12:28] exists and yet I still find myself
[12:32] spending time writing and rewriting
[12:34] things to tell beginners that I know
[12:38] would be helpful to them but are really
[12:41] boring for me to repeat.
[12:44] And so I've I've been writing
[12:45] non-fiction online for 15 years. I've
[12:48] been writing about how to write on the
[12:50] internet for over a decade. I have I'm
[12:53] not exaggerating like probably written I
[12:56] don't know north of 50 million words on
[13:00] the topic like thousands and thousands
[13:02] and thousands of pieces and especially
[13:04] in the past few years this frustration
[13:06] for me is like I know that there are
[13:10] still a tremendous amount of people in
[13:12] the world who need help taking the first
[13:14] step and going how do I start writing on
[13:16] the internet what's my niche how do I
[13:19] name and claim what makes me different
[13:21] ideas ideas that are very simple to me,
[13:23] that are very novel to beginners.
[13:26] And yet, every time I write about them
[13:28] for the 999th
[13:30] time, I feel like I'm wasting my finite
[13:34] time on Earth because I'm just
[13:36] re-wizzling the same groupings of words
[13:39] over and over and over again.
[13:42] And so, what kind of that that's what
[13:44] originally prompted this way of
[13:45] thinking. And it made me realize that
[13:48] commodity content doesn't have zero
[13:51] value. Commodity content does have
[13:53] value. And for a lot of people,
[13:55] particularly beginners in any industry,
[13:58] commodity content is actually the most
[14:00] helpful thing you could give them. The
[14:02] problem is that as you write more and
[14:05] more commodity content about a specific
[14:07] topic or in a niche, you will feel what
[14:10] I'm feeling. you probably already do
[14:12] where you're like, I find myself just
[14:14] repeating the same ideas that I've
[14:16] already thought through. I'm not
[14:17] actually thinking through anything new.
[14:19] I'm just repeating things I already
[14:20] know. And that is that is very low
[14:23] leverage. And so
[14:27] as you begin to build your content
[14:28] library,
[14:30] it you have to sort of examine why am I
[14:33] writing the things that I'm writing? And
[14:35] each new thing that I write, am I how
[14:37] what is the additive value
[14:40] of the thing that I just created to my
[14:44] what I'll call digital brain, my
[14:46] personal language model. And in an ideal
[14:49] world, like so just a quick little
[14:52] little backstory. Um I've been ghost
[14:54] writing for a decade. I used to do this
[14:58] framework with clients where when I
[15:01] would be working with a client who was
[15:02] really really busy, maybe they wrote a
[15:04] book, maybe they would have a podcast,
[15:06] more likely maybe they would just record
[15:08] internal meetings and I would pull from
[15:11] those assets, pull from their book or
[15:13] pull from blog post or pull from things
[15:15] that they had created or said out loud
[15:17] and I would take those words and phrases
[15:19] and I would assemble them into content.
[15:22] And that's essentially what a ghost
[15:23] writer does. The misunderstanding is
[15:26] that the ghostriter isn't really
[15:28] creating new ideas. The ghostriter is
[15:30] not really like putting net new thinking
[15:32] in the client's mouth. The ghostriter is
[15:35] essentially taking what I like to call
[15:37] approved language. The client said it.
[15:41] And so, why can't I take a sentence from
[15:44] chapter 1 and a sentence from chapter 2
[15:46] and two sentences from chapter 3 and two
[15:49] sentences from chapter 4 of their book
[15:51] and stitch them all together and create
[15:53] something new? The client said all of
[15:56] those words in those groupings of words.
[15:59] And so I did the manual version of what
[16:02] AI can and should do here soon, which is
[16:06] the larger your library grows, the the
[16:09] more ideas and approved language you
[16:12] have, the more net new combinations you
[16:14] can create. And I think that that
[16:17] fundamental idea is something that the
[16:19] masses do not understand about the
[16:22] leverage you get in crystallizing your
[16:24] thinking and building a library of
[16:25] content on the internet because the AI
[16:28] is not actually writing. AI is taking
[16:32] approved language from things that
[16:33] you've written and creating net new
[16:36] combinations. And and if you play that
[16:38] out, the benefit of doing that is now I
[16:41] I can continue to help beginners telling
[16:44] them the ideas I know they need to hear,
[16:47] but I don't need to take my finite time
[16:49] on Earth rewriting the same thing over
[16:52] and over again. Which what does that
[16:55] mean? That opens up more time for me to
[16:57] do what? To think about the next thing I
[17:00] haven't thought about. And I think this
[17:02] is the hope or this is the the utopian
[17:05] hope of AI is that AI will automate the
[17:08] lowlever stuff which opens up time for
[17:11] higher leverage things. And so in the
[17:14] case of writing, what that means is AI
[17:16] will help me repeat the same commodity
[17:18] ideas that I've already written 10,000
[17:20] times, which opens up time for me to
[17:22] think about non-commodity ideas,
[17:25] personality ideas, or ideally original
[17:27] ideas, which is more additive to the
[17:30] long-term value of my digital brain,
[17:32] which imagine the societal implications
[17:35] of more people thinking about things
[17:37] they haven't thought about before rather
[17:39] than just like turning out blog posts of
[17:41] things that they've already thought
[17:42] about and just need to repeat.
[17:44] >> To me, that creates two two bottlenecks
[17:48] and I'm curious your thoughts on this.
[17:49] The first is how do you create a second
[17:53] brain that you know people use Obsidian
[17:57] or whatever second brain software you
[17:59] want to use so that you're properly
[18:02] capturing those ideas, those ideas um
[18:06] are in a safe place. Um, and then the
[18:10] second bottleneck is okay, I've got
[18:12] these, you know, highquality ideas.
[18:14] Actually, some of them might be high
[18:15] quality, some might be lowquality, but
[18:17] I've got these ideas in this bank. How
[18:21] do I
[18:23] essentially plug it into an LLM such
[18:26] that I don't get slopp like there's
[18:28] there's a bunch of people who are just
[18:31] able to you know
[18:34] you know mangle their LLMs such that
[18:37] they understand
[18:39] how to uh prompt it in a way or or you
[18:43] know have agents run in a way such that
[18:46] it you know it's eloquent. So I think to
[18:49] me those are the two bottlenecks. Am I
[18:51] missing something?
[18:52] >> Yeah. I mean I think that there's
[18:54] there's certainly outlier use cases. And
[18:56] I'm I'm not against the sort of person
[18:59] who uh is not a writer, is not a
[19:01] proficient writer, doesn't want to be a
[19:03] proficient writer, but knows a lot about
[19:05] an industry, wants to brain dump into an
[19:08] LLM, and then have the LLM take that
[19:11] information and just write it the way
[19:13] that the LLM would. That's really no
[19:15] different than how ghostriters work with
[19:17] clients. And inherently in hiring a
[19:19] ghost writer, you get subjectivity,
[19:21] right? Ghostriters are human beings.
[19:23] They all write in different ways. You're
[19:25] going to get a subjective output because
[19:26] you're working with a specific ghost
[19:28] writer. An LLM is no different. And I'm
[19:31] not against that way of working. I think
[19:33] that the fundamental mistake and and the
[19:36] point of view that I'm really leaning
[19:37] into for the foreseeable future is that
[19:41] the highest leverage thing you can do is
[19:43] the writing. The robots should do
[19:46] everything else. And I think where where
[19:50] people get lost and why slop happens is
[19:54] they might agree or understand that
[19:56] idea, but then they want to skip the
[20:00] step of writing and go, "Well, my LLM
[20:02] knows enough and it can just do it for
[20:04] me." And you don't realize that in doing
[20:06] that, you're giving up a lot of
[20:07] leverage. I don't want the robots
[20:10] writing for me. What I want is the
[20:12] robots repeating me,
[20:15] >> which is different. So, for example, um
[20:19] you you post on a bunch of different
[20:21] social platforms, right?
[20:24] And I know that I'm like, "Okay, I have
[20:27] this giant library of short form text,
[20:30] giant. I've tweeted 10,000 times,
[20:32] whatever it is, right? And I want to
[20:34] keep publishing two short form pieces a
[20:37] day. And I also want to put out short
[20:38] form on Instagram threads and YouTube
[20:41] community posts and some of these other
[20:42] platforms.
[20:44] It is extremely low leverage for me to
[20:47] go through my library and find the
[20:50] highest performing posts and repost them
[20:52] on multiple platforms.
[20:54] People got around this with a social
[20:55] media manager. Cheaper solution, you get
[20:58] around that with a VA and even cheaper
[21:00] solution now AI and agents can
[21:02] essentially do that. But the step
[21:05] further and where where I see long-term
[21:07] value in this idea of building a digital
[21:10] brain or a personal language model is
[21:11] that it is not difficult for AI to
[21:15] understand. Uh Cole, you wrote this
[21:19] short form piece about uh five benefits
[21:22] of journaling and you wrote this other
[21:26] short form post on uh seven benefits of
[21:29] journaling. And I can take some from the
[21:33] first list and some from the second list
[21:34] and create a new third list that is just
[21:37] repeating a different combination of
[21:39] those two posts.
[21:41] And this this is where every time you
[21:43] sit down to write, I think one of the
[21:45] most helpful things to ask yourself is
[21:47] what is the short, medium, and long-term
[21:49] value of the thing that I'm creating?
[21:51] And I I'll give you an example of of
[21:54] like what so what is the value of you
[21:56] writing commodity content? If commodity
[21:58] content is a commodity, what's what's
[22:00] the point? The point is that
[22:04] there is borderline infinite commodity
[22:07] content inside every topic. The question
[22:10] is not um how do you write about every
[22:14] type of commodity content in your topic.
[22:16] The question is which types of commodity
[22:18] content do you agree with and which
[22:20] types do you disagree with. So I'll give
[22:23] you a simple example. In the world of
[22:24] personal finance, two pieces of
[22:27] commodity content are
[22:30] um if you want to build wealth, you
[22:32] should skip your morning coffee, don't
[22:34] waste that $3, and you should just buy
[22:36] the S&P 500 and 30 years from now,
[22:38] you'll be a millionaire.
[22:39] >> Can I guess?
[22:41] >> Yeah. Yeah.
[22:42] >> Cuz I'm just curious. Uh is the other
[22:44] one, if you only rented and didn't pay a
[22:48] mortgage, you would actually outperform
[22:51] like in general? Like I feel like I see
[22:52] that one. I I see that one all the time.
[22:55] >> Yeah. So, the these are examples of
[22:58] commodity ideas. These are not unique to
[23:00] any one individual, right? Like one
[23:03] person says, "Save your coffee, buy the
[23:05] S&P, you'll get rich." Another person
[23:06] says, "Saving your coffee money isn't
[23:08] going to make you rich. You're focused
[23:09] on the wrong thing." Both of those are
[23:11] commodity ideas. The question and why
[23:15] you should write about it is which one
[23:17] do you agree with?
[23:19] Because those are two different ways of
[23:20] looking at the world. And so where
[23:23] people run into, oh, AI is creating slop
[23:26] for me, is they don't realize how many
[23:28] of those micro decisions, they're
[23:30] expecting AI to just understand about
[23:33] them.
[23:34] >> So AI might, like, let's say their topic
[23:36] is personal finance. AI might write
[23:38] something and use the example, well, you
[23:40] should save your uh $3 of coffee money
[23:44] uh and just invest in the S&P 500. And
[23:46] the person reads that and goes, oh, this
[23:48] is slob. But they don't realize that the
[23:50] reason AI wrote that is because they
[23:52] don't know. AI doesn't know which
[23:54] perspective you agree with. So what is
[23:56] what is the long-term value of your
[23:58] library? Well, on the commodity content
[24:00] side, the long-term value is you
[24:02] actually yourself having clarity over
[24:05] which points of view do you agree with,
[24:07] which points of view do you disagree
[24:08] with. It's not the ideas themselves that
[24:10] are unique. It's which ones are tied to
[24:11] your identity. And the more of those
[24:14] that you crystallize, the more long-term
[24:16] leverage you get out of working with AI
[24:19] because AI knows, oh, I'm never going to
[24:20] use that example because that's not how
[24:22] you see the world.
[24:24] >> How do you actually like set this up?
[24:26] You know, if someone listening to this
[24:28] is like, yeah, I agree. I agree with
[24:30] Cole and what he's saying, you know, it
[24:32] makes a lot of sense. Um,
[24:35] how do I how do I actually set this up?
[24:38] >> Well, shameless plug. This is what this
[24:40] is the point of view. this is what we're
[24:42] we're building inside of TypeShare. So,
[24:44] if you want SAS that does that, there
[24:45] you go. But the reality is there's
[24:47] there's a lot of different ways that you
[24:49] can accomplish this. And fundamentally,
[24:51] it actually doesn't really matter where
[24:52] the library of content lives. What
[24:55] matters is that uh AI has access to it.
[24:58] And more importantly, like you can't
[25:01] skip the step of understanding what
[25:03] you're doing in the first place. I I
[25:05] fundamentally believe that the reason
[25:07] people are struggling with writing and
[25:10] AI slop and writing with AI is because
[25:14] deep down everyone wants to skip the
[25:17] step of thinking and just have the
[25:19] robots do the thinking for them. And in
[25:22] doing so, it's like you're trying to
[25:23] find a shortcut and the shortcut is the
[25:25] longer road in disguise. And so like an
[25:28] example that's very visceral for me is
[25:31] when I was writing on Quora back in 2013
[25:34] 2014. I wrote a handful of answers uh
[25:39] about me playing World of Warcraft
[25:41] professionally as a teenager. And I
[25:43] would always like in in each answer I
[25:45] sort of referenced it in different ways.
[25:46] I was like, "Oh, when I was a teenager I
[25:48] was a really competitive World of
[25:50] Warcraft player." And then another piece
[25:51] I would say the same thing just like a
[25:52] different combination of words. And one
[25:55] day I wrote a piece and the first
[25:57] sentence was when I was 17 years old I
[26:00] was one of the highest ranked World of
[26:02] Warcraft players in North America. And
[26:04] that Kora answer went viral. And so man
[26:08] like me myself manually looked at that
[26:11] and went that first sentence and that
[26:14] way that grouping of words that way of
[26:16] talking about that experience
[26:19] my hypothesis is clearly that resonated
[26:21] with people. So what did I do? I wrote
[26:26] hundreds of other pieces over the next 5
[26:30] 10 years repeating that exact sentence
[26:33] somewhere. Maybe it was the first
[26:35] sentence, maybe it was a hook somewhere
[26:36] in it, maybe it was before I told a
[26:37] story, but I used that grouping of words
[26:40] hundreds of times.
[26:42] That is the goal of what AI can and
[26:45] should help you leverage is that as you
[26:49] get clarity over these ideas and
[26:51] groupings of words and and points of
[26:52] view and perspectives, once you say it
[26:55] really well once or twice or two
[26:58] different variations and you like both,
[27:01] AI can then go, "Yeah, yeah, you've said
[27:02] that. I got it from here." You you want
[27:04] AI doing repeating. You don't want AI
[27:07] doing net new writing thinking. I'm
[27:11] assembling groupings of words and
[27:12] groupings of sentences that you've never
[27:14] done before. That's that's why then it
[27:17] gives you something and you're like,
[27:18] "This isn't me." Cuz it's not you.
[27:24] >> Yeah. And and people sniff it out, too.
[27:26] And it actually hurts you more than it
[27:29] helps you because then they associate
[27:32] you with some in my opinion then they
[27:35] associate you with someone who isn't
[27:38] adding value. And it's like there's
[27:40] nothing worse
[27:43] I don't think there's anything worse
[27:44] that you can do than steal someone's
[27:46] time. And I think that's why people are
[27:48] so you know hate slop so much. It's It's
[27:51] cuz they consume this piece of content
[27:53] and they're like, "Wow, that was sloppy.
[27:56] I hated that."
[27:58] >> Mhm. I I think this is a a second really
[28:00] fascinating question in the age of AI,
[28:03] which is so what is ownership? If if
[28:07] language is open-source, no one owns the
[28:10] alphabet. No one owns 99.9% of the
[28:14] words, and most people don't even own
[28:15] 99.9%
[28:17] of the groupings of words. If language
[28:20] is open source, what is ownership?
[28:24] And ownership um and and obviously
[28:27] there's like a legal component. It's
[28:28] like you can trademark names and like
[28:31] there there is a legal component, but
[28:33] generally speaking, ownership is really
[28:35] just a a different word for association.
[28:40] You associate ideas with certain people.
[28:44] Now, you can achieve association in lots
[28:46] of different ways. The easiest way to
[28:48] achieve association is to write about a
[28:51] topic a lot.
[28:54] Like people associate AI startup ideas
[28:58] with you. And we can just say at a
[29:00] fundamental level for no other reason
[29:02] than you create more about that topic
[29:05] than the vast majority of people.
[29:07] >> Or I'll give you I'll give you another
[29:08] example two examples of your coffee uh
[29:12] drinking coffee and wasting money which
[29:14] is when I hear
[29:17] You know that concept I think of Ramit
[29:21] Si I will teach you to be rich who
[29:23] always says keep drinking your coffee
[29:25] even if it cost $6 like live your rich
[29:28] life and that's against the common idea
[29:32] of you know save put in the S&P and the
[29:35] other side of that coin is Dave Ramsey
[29:38] who's like save invest don't go in debt
[29:42] and so you you know both ends of the
[29:44] spectrum same core idea.
[29:48] >> Exactly. And I mean, so to build on
[29:51] that, right, Dave Ramsey is known for
[29:53] that meme where he's like taking the
[29:55] live call and he's like, "So you spend
[29:57] all your money on what?" And then it's a
[29:59] funny answer and he's like, "And you owe
[30:01] how much money to who?" And then it's a
[30:03] funny answer. Like what's actually
[30:05] happening is that that's a format that's
[30:07] a structure that people associate with
[30:12] him. So o ownership is really just
[30:16] association
[30:18] uh generally speaking and the easiest
[30:21] way to establish association is through
[30:24] volume. You write a lot about a topic,
[30:26] people associate that topic with you.
[30:28] Very basic. The the next layer up of
[30:32] association is um personality.
[30:36] It is okay. So, it's not just that you
[30:39] write about this topic, but the audience
[30:42] knows all of these specific things about
[30:44] you. They know like so, uh, play it out.
[30:48] It's like you have a, um,
[30:51] like a personal finance expert and you
[30:53] know nothing about them. They just write
[30:54] a lot about personal finance and then
[30:56] there's another personal finance expert
[30:58] who is maybe just as qualified or less
[31:00] qualified, but they share a lot about
[31:02] their personal life, their personality,
[31:04] how they see the world. She's a personal
[31:06] finance expert. She's a single mom. She
[31:09] got divorced three years ago. She has
[31:11] two kids. She's had to work multiple
[31:13] jobs to put them both, you know, through
[31:14] school. Uh really hard worker. Started a
[31:18] blog. Blog took off.
[31:20] You know, all these things about that
[31:22] person. And something that I think a lot
[31:25] of people forget is that quote unquote
[31:29] less credible
[31:32] more personality content typically
[31:35] outperforms
[31:37] more credible commodity content. So like
[31:42] take take the extreme like there are
[31:44] there are not as successful personal
[31:47] finance people or finance people with
[31:50] significant personality content that get
[31:53] more distribution and have bigger
[31:55] audiences than Ray Dalio
[31:58] who's like a multi-billionaire
[32:00] right so the credibility of the person
[32:03] is actually not like it's important you
[32:05] want to learn or or hear from someone
[32:07] who has done something and you know
[32:08] knows about the
[32:10] But I think people forget that
[32:11] credibility itself is not the the the
[32:14] winning card. The winning card is
[32:17] actually the personality of the person
[32:18] and how much they're willing to let you
[32:21] into their world and how many things
[32:22] about them you associate with that
[32:25] topic. And that that idea should be
[32:28] really liberating for a lot of people.
[32:30] Like I'm sure a lot of your listeners,
[32:31] people that build products or are
[32:33] first-time founders or second time
[32:34] founders, they want to build organic
[32:36] audiences. Like you actually don't need
[32:39] to be coming up with gamechanging,
[32:41] radically new, original thinking. All
[32:44] you need to do is just go, "Yeah,
[32:46] there's a gazillion startup founders,
[32:47] but I am a startup founder, and these
[32:49] are this is what makes me unique. This
[32:51] is how I see the world. This is where I
[32:53] live. This is how many siblings I have.
[32:55] This is where I went to school. This is
[32:56] what I studied. This is what my hobbies
[32:58] are." And all of those little data
[33:00] points actually now in the age of
[33:03] infinite textbased commodity content,
[33:05] all of those little data points are
[33:06] differentiators. And more importantly,
[33:09] every time you write about one of those
[33:11] personality details, that personality
[33:14] detail should make its way into your
[33:16] digital brain, your personal language
[33:17] model. Just like with me, when I was 17
[33:20] years old, I was one of the highest
[33:21] ranked World of Warcraft players in
[33:23] North America. There's a reason why I
[33:25] repeated that 500 times, you know. And
[33:28] so then association is really just I
[33:30] associate these uh life life details,
[33:33] life events, ways of seeing the world
[33:36] with you. Like I'm sure you're familiar
[33:38] with um Gary Vee repeated this has been
[33:41] repeating this for 15 years, which is I
[33:43] want to buy the New York Jets.
[33:45] >> Mhm. O
[33:47] >> over and over and over and over and over
[33:49] again. Like what is that? That's a
[33:50] personality detail. And you associate
[33:52] that detail with him. And I think he
[33:54] lives in New Jersey. [laughter]
[33:57] [gasps] So, it's kind of funny, right? I
[33:59] know he works maybe he has a a place in
[34:02] Manhattan, but you know, I think he grew
[34:04] up and and still lives in New Jersey.
[34:07] So, it is interesting that you can have
[34:08] someone who lives in another place and
[34:10] his dream is to buy the the New York
[34:13] Jets and, you know, tens of millions of
[34:15] people associate Gary Vee who's not, you
[34:18] know, necessarily a sports person. And I
[34:21] know he has a sports business now, but
[34:23] you know, made his wealth in in wine and
[34:25] and and agencies, social agencies. All
[34:29] of a sudden, people connect him with uh
[34:32] with the New York Jets. I think what's
[34:34] what's something really interesting you
[34:36] said is
[34:38] is basically like, hey, you're going to
[34:40] have your competition basically is going
[34:43] to just plop ideas to create commod
[34:47] commoditized content and there's just
[34:50] going to be more and more of that.
[34:52] Whereas if you could do a audit of of
[34:58] interesting tidbits about you, your
[35:00] brand, how you think about things, point
[35:02] of views as you call it, um throw it in
[35:04] a markdown file or use a second brain
[35:07] software like Obsidian or something, and
[35:12] it's going to just stand out by virtue
[35:15] of everyone else is is is using the same
[35:19] tool and the same methodology to crank
[35:21] crank out content. So, it's like it's
[35:24] it's just it's sort of like fish fish
[35:27] where the fish are and the fish. You're
[35:31] going to be able to fish with dynamite
[35:33] if you actually have these stories and
[35:35] stuff like that.
[35:36] >> Yeah. To to sharpen the precision of the
[35:40] language, it's that everyone is using
[35:43] the same LLMs, meaning they are all
[35:46] creating from the same data set.
[35:48] >> Yeah. your your personality content,
[35:51] your library of content is the unmade
[35:54] data set. You can't you can't expect AI
[35:58] to write a paragraph inside of an essay
[36:02] or some or a newsletter or something
[36:04] that you're writing about what you
[36:06] learned as a kid going through an
[36:07] experience that you've never
[36:09] articulated.
[36:11] AI doesn't know that. That is an unmade
[36:12] data set. And so the way I think about
[36:15] personality content and that that aspect
[36:17] of a personal language model is that you
[36:20] want to spend your time building the
[36:22] unmade data set. And the way that you do
[36:24] that is by writing about how you became
[36:26] you. Like what are your formative
[36:28] experiences? What lessons did you learn?
[36:30] What mistakes did you make? The all of
[36:32] these things seem very mundane to
[36:36] ourselves, but that's because we lived
[36:38] it. We experienced it. We take it for
[36:40] granted. But you articulating a a
[36:42] visceral mistake, especially in a story
[36:45] instantly differentiates that piece of
[36:47] writing from everybody else writing
[36:49] about the [clears throat] same
[36:50] fundamental commodity idea, right? And
[36:53] so that's where you want your time to be
[36:55] spent is you you want to you want to be
[36:57] building these net new additive data
[36:59] points, the unmade data set about how
[37:01] you became you. What do you say to
[37:03] people who are thinking to themselves,
[37:06] well, I just I lead an uninteresting
[37:08] life and I don't have anything
[37:11] interesting, you know, or specific or
[37:13] unique about me.
[37:15] >> Yeah. I mean, I a that's incorrect. Um,
[37:19] and b, like I don't know any other way
[37:21] to say it. It's just wrong. Um, and and
[37:24] B, it's like you have to you have to get
[37:27] outside of yourself and realize that
[37:29] even the most mundane experiences are
[37:31] things that other people resonate with
[37:34] and see themselves in. Like a really
[37:36] simple example is I moved across the
[37:38] country twice in the past four years.
[37:41] That's not like the most like unique and
[37:44] different and captivating data point
[37:46] ever, but it it is an experience. And I
[37:49] bet if I was at a dinner party and I
[37:50] said, "Yeah, I moved coast to coast
[37:52] twice in the past four years." I might
[37:55] be talking with someone else. And
[37:57] they're like, "Oh my gosh, I did the
[37:58] same thing 6 years ago. Wasn't it
[38:00] crazy?" And then we're telling each
[38:01] other stories and all of a sudden we're
[38:02] like friends at a dinner party. We we
[38:04] fail to realize that we do all of these
[38:06] things in conversation every single day
[38:09] of our lives. and the the the build- on
[38:13] idea. And I I've been thinking about
[38:15] this for a decade because I I was trying
[38:18] to explain this to ghost writing clients
[38:19] back in 2016 2017, which is in real
[38:24] life,
[38:26] we walk around as [clears throat] a
[38:28] walking talking personal language model
[38:32] already. And what happens is we go have
[38:35] a coffee meeting. Like think about the
[38:37] first time we we met up for coffee in
[38:39] Miami. and you you go have a coffee
[38:42] meeting or or meet up with someone and
[38:44] what do you do? You start sharing things
[38:47] from your personal language model.
[38:49] You're like, "Here's where I'm from. Uh
[38:51] here's how many siblings I have. I moved
[38:53] to this city 2 years ago. Here's what
[38:55] I'm into. Oh, you're into the same
[38:57] thing." We're just doing the unscalable
[39:00] version of that. Writing on the internet
[39:03] is the idea that instead of repeating
[39:05] the same life facts and stories and
[39:08] perspectives over and over again
[39:10] manually at one meeting and then another
[39:13] dinner party and then another grouping
[39:14] with friends and then a picnic one
[39:16] afternoon. Instead of doing that
[39:18] manually, what you can do is you can
[39:20] write about it once and then publish it
[39:22] on the internet and have that same
[39:24] coffee meeting sharing that same
[39:26] perspective or insight or life fact or
[39:28] story with borderline infinite people.
[39:31] And then with AI, that piece now gets
[39:34] added into your library. And now you can
[39:36] go, I don't want to I don't want to tell
[39:38] that story anymore. I've already told
[39:40] that story 10 times. I've written about
[39:41] it 10 times. We've got 10 different
[39:42] variations. I just want you to repeat it
[39:44] for me. And that's that is the
[39:47] fundamental leverage of writing with AI
[39:50] in for for the next decade, foreseeable
[39:53] future.
[39:54] >> So that works with personal brand. Like
[39:56] if you're trying to build a personal
[39:57] brand, you want to get out there, you
[39:59] want to be Gary Vee, that works. How
[40:02] does this work with branded content?
[40:07] First of all, I think that
[40:10] for a long time, I remember my my first
[40:12] job out of college was working at an ad
[40:14] agency, and I was repeating this as a
[40:17] 23-year-old. I was like trying to tell
[40:19] my boss and my boss's boss like, "Nobody
[40:22] wants to hear from the branded Twitter
[40:24] page. They want to hear from the CEO.
[40:26] They want to hear from a a human being.
[40:29] And that only proves more and more true.
[40:31] So I think that regardless the the human
[40:34] being personal branding decision is
[40:37] continuing to accelerate and branded
[40:39] content just always underperforms.
[40:42] But also there's no reason why your
[40:45] branded content can't repeat all of your
[40:47] chosen commodity ideas. like if we
[40:50] create so let's let's say Typesh Share
[40:52] our SAS platform has uh an X account and
[40:55] a LinkedIn page
[40:57] there's absolutely no reason why those
[41:00] social pages can't just repeat all of my
[41:02] commodity ideas now it wouldn't make
[41:03] sense for Typeshare to be like when I
[41:05] was 17 years old I was one of the
[41:07] highest ranked World of Warcraft players
[41:08] in North America like that doesn't make
[41:09] sense because that's a personality data
[41:11] point but there's no reason why
[41:13] Typeshare can't tweet here are the five
[41:15] benefits of writing online it's a
[41:18] commodity and it's the same thing I
[41:19] would have said. So I actually I
[41:21] actually think that you get double
[41:23] leverage when you have the human being,
[41:26] the the founder or an executive team or
[41:28] whoever building their own personal
[41:31] language model that then the branded
[41:33] content is like great, we're just going
[41:34] to ignore the personality details and
[41:35] we're just going to repeat the commodity
[41:36] ideas, which is what brands do anyways.
[41:39] You're just choosing which perspectives
[41:41] do we agree with.
[41:43] >> Yeah. And I I think you can still, you
[41:45] know, there's so many brands, there are
[41:47] a lot of brands like Taco Bell comes top
[41:50] of mind, which like it feels like it's
[41:52] personality driven.
[41:54] >> So like just because it's commoditiz
[41:57] commoditized content doesn't mean it
[42:00] can't
[42:01] be human centric. like you can have like
[42:05] I think the future of branded content is
[42:07] like
[42:09] you know basically not just one human
[42:12] behind it but like five you know and you
[42:15] can even like sign off oh it's Tiffany
[42:17] who who did this or Jeremy who did this
[42:19] who wrote this and different stories
[42:22] around the brand so for example like you
[42:25] know the Taco Bell let's just use that
[42:28] like you can have instead of just Taco
[42:30] Bell writing commoditized content around
[42:32] or just content around oh hey like we
[42:35] dropped like a new crunchy burrito today
[42:37] and here's what it looks like. It's more
[42:39] around um you know Tiffany who works at
[42:43] the you know Tulsa Oklahoma store used
[42:47] has been coming to Taco Bell for the
[42:48] last since she's been 13. In fact her
[42:51] parents met at a Taco Bell
[42:54] >> and then you you tell that story.
[42:57] >> Think about Okay, so that's a fantastic
[42:59] example. Think about how
[43:01] with a little bit of creativity, if you
[43:03] take or Taco Bell takes that approach,
[43:06] what what's really happening is you're
[43:09] you're opening the door for personality
[43:12] content. And personality content, that's
[43:15] that's what I was saying where where
[43:17] lots of like high personality content
[43:20] often outperforms more credible
[43:22] commodity content. And so I actually
[43:25] think in that example, Tiffany, the
[43:27] tacoar
[43:29] would outperform, right, the other the
[43:32] other taco company brand that's just
[43:35] only doing the commodity version. And
[43:38] this this this is why I think this lens
[43:40] is so helpful is because it it forces
[43:43] you to raise the question like what are
[43:45] we actually creating here? What is the
[43:47] value? And if you're only creating
[43:49] commodity content with no personality
[43:51] component, you have to fundamentally
[43:53] realize it's like you are creating the
[43:55] lowest leverage lower lowest value form
[43:57] of content in your topic. That's not to
[43:59] say it has zero value, but it has the
[44:01] lowest value. A little a little
[44:04] creativity means you can open the second
[44:05] trunch which is well now we give
[44:07] ourselves permission to create some
[44:09] personality content. And the ultimate
[44:12] goal, the ultimate goal is to spend
[44:14] enough time crystallizing your thinking
[44:17] through writing on a specific topic
[44:20] where you ultimately create original
[44:23] content. And I one one of the things
[44:26] that I've been writing about recently is
[44:29] just because you wrote it doesn't mean
[44:30] you own it.
[44:32] I think that a lot of times people think
[44:35] because my fingers touched the keys, my
[44:39] fingers typed that grouping of words,
[44:41] that means that grouping of words is
[44:42] mine. And that's not true. If you write
[44:45] the sentence, I woke up today and made
[44:47] coffee.
[44:49] That is not attributable to you. That is
[44:51] an open-source grouping of words. No one
[44:53] owns that. And and there's there's this
[44:57] uh faulty belief that people have where
[44:59] they think if they're the ones who wrote
[45:02] it, they're the ones who did the work,
[45:04] that means they own it, and that means
[45:06] they created something original.
[45:09] That is not original thinking. You
[45:11] didn't you didn't think a single new
[45:13] thought. You didn't say anything that
[45:15] hasn't been said before. And so original
[45:18] thinking is not something that's like,
[45:20] oh, I started writing online and 3 days
[45:22] later I had a game-changing insight.
[45:24] Original thinking comes as the result of
[45:26] crystallizing the way that you see the
[45:29] world in a specific topic or a small
[45:31] number of topics for a prolonged period
[45:33] of time. And one of the best examples of
[45:36] original thinking like original does not
[45:38] mean I wrote a 100,000word book on it.
[45:41] Great example of original thinking was
[45:43] Pythagoras saying the earth is round not
[45:46] flat.
[45:47] That's six words, like fundamentally six
[45:51] words that completely changed the course
[45:52] of humanity that we still associate with
[45:55] that person.
[45:57] And you only arrive at those six words
[46:00] after
[46:01] years and years and years of
[46:03] crystallizing thinking in a topic. And
[46:05] so if you kind of loop all the way back
[46:07] to the beginning here, that's why
[46:08] there's there's an irony when people
[46:11] say, "I want to write about everything."
[46:14] It's like you don't realize that in
[46:15] doing that you deprive yourself of ever
[46:18] being able to go deep enough where you
[46:20] end up saying something original. All
[46:22] you're doing is you're just typing and
[46:24] repeating commodity ideas all day long
[46:26] thinking they're yours and thinking that
[46:28] you have ownership or originality
[46:30] because you were the one who wrote the
[46:31] words when in reality those groupings of
[46:33] words have existed forever.
[46:35] >> How do you see writing versus short form
[46:39] video?
[46:41] >> In what sense? So we talked a lot about
[46:45] content in general or on you know
[46:48] capturing ideas, writing ideas and when
[46:51] I hear that I think of
[46:54] uh tweets, I think of blog posts, I
[46:56] think of email newsletters, I think of
[46:59] copy on websites.
[47:01] But on the other hand, we have uh all
[47:04] the social feeds just hungry for short
[47:07] form video. So my question to you is how
[47:10] do you think about writing as it relates
[47:13] to short form video?
[47:15] >> First of all, I don't One of the most
[47:18] common questions people ask is which is
[47:20] better? Which is better, writing or
[47:22] video? Which is better, podcast or long
[47:24] form YouTube? The reality is that the
[47:27] answer is yes. It's all of them. They
[47:30] all work.
[47:32] Me and Dicki, my business partner,
[47:34] create
[47:36] essentially zero short form video. The
[47:39] only the only short form video we
[47:41] publish is we have a uh internal editor
[47:44] and we record a long form and he clips
[47:46] out some random short form and posts
[47:47] them. Like it is like very low, I'll
[47:51] even say low value short form. We are
[47:53] not like intentionally investing in
[47:55] that. And we run a multi-sefigure
[47:57] business. It's all driven through
[48:00] writing. So like the yes, there's always
[48:04] a trend or some other form of content
[48:06] that is working or maybe working a
[48:09] little bit better, but that doesn't mean
[48:10] that the other ones aren't working. You
[48:12] know, this the second idea is
[48:16] um video is no different than writing.
[48:18] Everyone uses teleprompterss now. Like
[48:20] when you do shorts, do you ever script
[48:22] them?
[48:24] >> I I'm I actually do. I script them when
[48:28] I do shorts. Uh, it's like cutouts of
[48:33] long form content
[48:36] or it's um I do like animation
[48:40] explainers that are scripted.
[48:42] >> So they are scripted.
[48:45] >> Yeah, those are scripted. Either it's
[48:47] cuts or scripted.
[48:49] >> Yeah.
[48:49] >> Yeah. Mostly.
[48:50] >> So So more and more short form video is
[48:53] scripted and a script is what? a script
[48:55] is writing and it's the same fundamental
[48:57] idea to to make this really tangible
[49:01] like walking through the the value of
[49:03] the thing that you're creating. So
[49:05] pretend I create a short form video and
[49:07] the first sentence is um here's how to
[49:11] start a newsletter on Substack. That is
[49:15] commodity content. That's not to say
[49:17] that it's not valuable, but that is
[49:20] fundamentally commodity content because
[49:22] I am just explaining how to do something
[49:23] that a lot of other people can and
[49:26] probably have explained how to do. If we
[49:28] change the first sentence and we say, I
[49:31] run a $400,000 a year annually annual
[49:34] recurring revenue paid newsletter on
[49:37] Substack. Here's how to start a
[49:39] Substack. Now, we've moved up a tunch
[49:42] that has a personality detail in It's a
[49:44] little like credibility too, but that's
[49:46] really a personality detail. Like I did
[49:48] this, I'm doing this, right? If we add
[49:51] in the variable, which is sort of the
[49:53] last variable we haven't talked about,
[49:55] which is timely versus timeless,
[49:57] everything that you create sits on a
[49:59] spectrum. Timely content has very high
[50:02] short-term value and very low long-term
[50:05] value.
[50:07] Um, timeless content has very low
[50:09] short-term value and very high uh
[50:12] long-term value. And so, for example, I
[50:14] saw you tweet um yesterday or two days
[50:16] ago like a like a AI image meme where
[50:19] you were like, "I'm not waiting for the
[50:21] foldable phone. I'm waiting for the the
[50:22] desktop that has that's foldable."
[50:24] Right? So, that's a great example of
[50:28] hyper timely content where you
[50:30] capitalize for 24 hours, but then the
[50:33] value of that content falls off a cliff.
[50:35] You you can't repost that a month from
[50:37] now or 3 months from now or a year from
[50:39] now, right? So this raises a very
[50:42] interesting question of all right so
[50:44] everything that you create like how much
[50:46] value are you actually extracting and
[50:48] over what time horizon you created a
[50:50] meme you capitalized for 24 hours great
[50:53] but like that's all the juice you're
[50:55] going to get out of it you're not going
[50:56] to get any long-term value out of that
[50:59] in contrast I wrote this essay about um
[51:01] AI took me two weeks to write spent a
[51:04] lot of time on it um it got some
[51:07] short-term value but I'm going to
[51:09] reference and use that essay for like
[51:11] the next 10 years.
[51:13] And so when you're creating something,
[51:15] you want to ask, well, where on the
[51:16] spectrum am I sitting? And are my
[51:19] expectations aligned with the thing that
[51:21] I'm creating? You can't go build
[51:23] something or write something or create
[51:25] something that, you know, is intended to
[51:27] be very timeless content and expect it
[51:29] to get the same short-term distribution
[51:31] as a meme that capitalizes on an Apple
[51:33] announcement in 24 hours, right? They
[51:36] there are different expectations. So, if
[51:38] we tweak the first sentence of that
[51:40] short form video and we go, um, Substack
[51:43] just launched this new feature. I run a
[51:48] $400,000 a year paid newsletter. Here's
[51:51] how to start a Substack in 2026.
[51:55] That word, this just happened, or in
[51:57] 2026,
[51:59] fundamentally changes the short, medium,
[52:02] and long-term value of that content
[52:04] because I can't repost that in 2030.
[52:07] Right? I maybe I could use pieces of it
[52:09] or I got to re-record it or I got to
[52:11] rewrite half of it. And so when you when
[52:13] you put all these variables together,
[52:16] you start to realize like everything you
[52:18] create in some way is is adding
[52:23] something to your digital brain, your
[52:25] knowledge of the world, your personal
[52:27] language model. The question is how much
[52:30] did it add? How additive was it? If if
[52:33] you spend 20 minutes and you're just
[52:34] repeating the same commodity idea that
[52:36] you've repeated a hundred times, maybe
[52:39] you added like one point of I said it
[52:41] slightly differently. Okay, cool. We
[52:42] have another like grouping of words
[52:44] approved language. Whereas if you sit
[52:47] down and write a net new essay about a
[52:49] topic you've never written about before
[52:50] and you crystallize a whole point of
[52:52] view around it, the additive value of of
[52:55] that to your digital brain is
[52:56] exponentially higher than the first one.
[53:00] And so that's that's my hope with AI is
[53:02] that the human is doing the writing, the
[53:04] human's doing the the thinking, the
[53:06] human is doing the choosing of how do I
[53:08] see the world, what ideas do I agree
[53:09] with, do I disagree with, what are my
[53:11] personality details, etc. But the robots
[53:15] are very effective at repeating and
[53:19] remixing and doing all the things that
[53:21] social media managers and even ghost
[53:23] writers and VAS have done for a very
[53:26] long time.
[53:27] You still got to give the robots
[53:29] direction around
[53:32] writing in a style that isn't generic
[53:35] sounding AI.
[53:38] So now you you have touched on the I
[53:41] think the final piece and it is the
[53:42] hardest one for people to understand
[53:44] which is voice. So what is voice? What
[53:46] is style?
[53:48] Voice is a bundled term. When when we
[53:51] say some someone has a voice, someone's
[53:53] content has voice, um we think that it's
[53:56] this intangible quality that can't be
[53:59] replicated, the reality is voice is
[54:01] really a bundled term made up of a
[54:03] couple different data points. It's like
[54:06] word choice dictates voice, right? If I
[54:09] say something is really cool, that gives
[54:12] you one perspective of who I am. If I
[54:15] say something is exquisite, that gives
[54:17] you a different perspective of who I am.
[54:19] And if I say something is dope,
[54:21] that gives you a different perspective
[54:23] of who I am. So individual word choice,
[54:26] like literally a single word in a
[54:27] sentence can change your perspective of
[54:30] what the voice of that sentence or
[54:32] person is. Another one is is sentence
[54:35] length and sentence structure. If
[54:37] someone writes a paragraph with no
[54:40] periods and like seven different
[54:43] sentences broken up by semicolons,
[54:46] you have a very different perspective of
[54:48] that person rather than someone who
[54:49] writes in short tur sentences with very
[54:53] simple third grade reading level
[54:55] vocabulary.
[54:57] And what's what's important about this
[54:59] is your your takeaway should be that
[55:02] voice is something that you
[55:03] intentionally create. It's not something
[55:06] that just naturally happens.
[55:08] And when you become proficient enough at
[55:11] writing,
[55:12] what's funny, I ran into the same thing
[55:14] ghost writing for for clients. I will
[55:16] never forget um I had a ghostriting
[55:18] client who I sent him a piece and he
[55:21] sent it back and he was furious. And he
[55:23] said, "This doesn't sound like me." And
[55:26] I looked at it and I went back and
[55:28] listened to the call and I was like,
[55:28] "All right." He didn't really tell me
[55:30] what it was, so I have to guess what
[55:31] he's thinking. Like, "Why doesn't this
[55:32] sound like him?" And it occurred to me,
[55:34] oh, I was I was using the wrong voice. I
[55:37] I made this piece uh lean too much into
[55:40] opinion. And this guy actually spoke
[55:42] referencing like a lot of studies and a
[55:43] lot of facts.
[55:45] I didn't rewrite the piece. All I did
[55:48] was go back in and add in like two or
[55:51] three sentences that just said according
[55:53] to the New York Times or according to
[55:55] Harvard Business Review. I changed 1% of
[55:58] it. And I sent it back to him and he was
[56:01] like elated and he was like, "Oh my
[56:02] gosh, this is so much better. It sounds
[56:04] just like me." And so the irony of
[56:06] writing with AI is often times AI will
[56:08] give you an output and people will look
[56:11] at one single variable. They'll look at
[56:13] one individual word choice and go the
[56:15] whole thing is And the reality is
[56:18] it's like no. If you were proficient,
[56:20] you understood the qualities that made
[56:22] your voice your voice, which most people
[56:23] don't know. All you would have to do is
[56:26] just go, I want to trade that one
[56:28] adjective for a different adjective. And
[56:30] all of a sudden, the sentence sounds
[56:31] like me.
[56:32] >> What do you think of people who don't
[56:35] like their voice, but who know which
[56:40] creators voices they like. So, they
[56:42] might say, you know, I don't really like
[56:43] my voice, but I really like how Greg
[56:46] Eisenberg writes his essays on X.
[56:49] >> I I truly believe that uh 99% of the
[56:53] time when someone says that, they can
[56:56] neither articulate what makes their
[56:58] voice their voice, what it is they don't
[57:00] like, and they can't articulate what
[57:02] makes your voice your voice. And usually
[57:05] when people say that, it's more of a
[57:07] mask of they're just afraid to write.
[57:10] Period.
[57:11] And so it's a lot easier to point to and
[57:13] make this like a voice issue. The
[57:15] reality I have ghost written for north
[57:17] of 300 different founders, executives,
[57:22] investors. The vast majority of the time
[57:25] I used the exact same voice in the
[57:27] writing. It was authoritative. It was
[57:30] concise. it it maybe had like a couple
[57:33] little vocabulary choices in there
[57:35] specific to that person, but like 98% of
[57:38] the thing was just an authoritative
[57:40] thought leadership style voice. And
[57:42] almost every one of them said, "This
[57:44] sounds like me." And so if you go if you
[57:47] go I'm not I'm not joking. And so if if
[57:49] you go back to like language is open
[57:53] source, most groupings of words are
[57:55] commodities. The reality is that the
[57:58] voice component in writing is a minority
[58:00] component. It's a very small piece. It's
[58:04] not like this entire piece of writing is
[58:06] an original voice. Very rarely is that
[58:08] the case. It happens, but very rarely.
[58:11] Instead, it like a piece is 80%
[58:13] commodity groupings of words with maybe
[58:16] some subjective swapping of adjectives
[58:19] and things. And I think this is one of
[58:21] those it's like you can't possibly
[58:24] discover the qualities that you want or
[58:26] like about your voice without writing.
[58:31] How are you going to discover, oh, I
[58:33] actually prefer these types of
[58:34] adjectives over these types of
[58:35] adjectives? You know, I prefer this this
[58:38] rhythm in my writing opposed to this
[58:40] rhythm in my writing. And so if if
[58:42] you're not willing to do that, h how in
[58:44] the world should you expect AI to have
[58:48] any inkling of any of those data points
[58:50] or choices?
[58:52] That's the unmade data set.
[58:54] >> Yeah. I mean, the truth is, you know,
[58:56] like deep down, you know, like it's the
[58:58] same it's the same concept around like
[59:00] you know what kind of food you like. You
[59:01] know if you're a pizza or a pasta
[59:03] person. And so like of course you know
[59:06] the types of words that you connect with
[59:08] how you how how you know your voice.
[59:12] It's just it it might be a little deep
[59:13] down you have to you have to be a little
[59:16] self-reflective on it but it it's there.
[59:19] And this is not to be clear this is not
[59:21] exclusive to the everyday person. I
[59:25] think oftenimes people think you know oh
[59:27] I'm I'm the everyday person. I haven't
[59:29] achieved this massive outcome. Um, I bet
[59:31] this is so much easier for people who
[59:33] have achieved all these things, right?
[59:35] They they make up all of these problems.
[59:37] The reality is I have I have ghost
[59:40] written for some of the most like you
[59:43] can't accomplish things any more
[59:44] accomplished than what they've
[59:45] accomplished in the world of business.
[59:48] And even they the core problem was
[59:54] they they get nervous staring at a blank
[59:56] page and they're they're nervous to put
[59:57] themselves out there.
[59:59] The majority of my value as a ghostriter
[60:01] was not actually the writing. It was
[60:03] helping the client and give giving them
[60:06] permission to put themselves out there.
[60:09] And it was comical to me ghostriting for
[60:11] so many people where I would have a
[60:13] client be like, I'm really afraid of
[60:15] putting my perspective out there. And
[60:16] their perspective is like the key to
[60:18] marketing is knowing your customer. I'm
[60:20] like, there's nothing there's nothing to
[60:23] be afraid of. You are literally
[60:24] repeating a commodity idea that most of
[60:26] the planet agrees with. Like and so a
[60:29] lot of these are self-inflicted issues
[60:32] and the the whole key is like getting
[60:35] out of your own way and hitting publish
[60:37] so that you can discover what these
[60:41] variables are. You know what makes your
[60:43] personality, what makes your writing
[60:44] style, what makes your voice.
[60:47] >> The other piece of that is generally
[60:50] speaking people are underpublishing.
[60:53] like the volume of content that they're
[60:55] putting out is lower than
[60:59] uh than they can actually do. I was
[61:01] having this conversation with um our
[61:05] agency actually and you know LCA like
[61:10] literally works with the biggest
[61:12] companies on the planet building
[61:15] their new products. So there's as you
[61:17] can imagine in the AI world as you can
[61:19] imagine you know that team has just an
[61:22] incredible uh you know data set and
[61:26] incredible knowledge base around what's
[61:29] working and what's not and there's so
[61:30] much content that they could be putting
[61:32] out and they put out some content here
[61:35] and there but it's I you know I think
[61:37] one of the things we're talking about is
[61:39] like h how do we ship a minimum of five
[61:42] pieces of content a day and what does
[61:44] that look like? What systems do we need
[61:46] in place? And I, you know, we could use
[61:49] AI in in that process, but what systems
[61:51] do we need in place for that to happen?
[61:53] Cuz if we're not posting five pieces of
[61:55] content per day, I don't believe we're
[61:58] giving it a fair shot. I would agree and
[62:01] [clears throat]
[62:02] I would say that it's important for
[62:05] people to remember that that is in many
[62:07] ways that is the baseline expectation
[62:09] now. And that is just the
[62:12] infrastructure. the the the question on
[62:14] top of that is what is the input? If
[62:16] you're posting five times a day, like
[62:18] where are you pulling the ideas from?
[62:20] Goes back to well, you need a digital
[62:22] brain, you need a personal language
[62:23] model, like where where is it all coming
[62:25] from? And the same way I'm I know you're
[62:28] aware of the trend. You know, you see uh
[62:30] like OpenAI acquired that podcast DBPN
[62:34] and um there's a huge move to these like
[62:38] large companies realizing they need a
[62:39] media component. I think that it might
[62:43] be idealistic, but I think that some of
[62:46] the best marketing money that any
[62:48] Fortune 500 company could spend would be
[62:51] plucking really thoughtful journalists
[62:56] and non-fiction writers in relevant or
[63:00] tangentally relevant fields and being
[63:02] like, you have one job. We're gonna pay
[63:04] you a fat salary and your only job is to
[63:08] think deeply about our industry and
[63:10] write long form content about it. That's
[63:13] it. Because what's really happening is
[63:16] it's not that like that individual piece
[63:18] of writing is the silver bullet and you
[63:20] know that's going to drive all the sales
[63:22] and they make it all super promotional.
[63:23] What's really happening is you have a
[63:25] very smart person who's paid to think
[63:28] deeply about a topic, hopefully over
[63:31] time creating net new original thinking
[63:33] and ideas that get associated with the
[63:35] company. And that library of content
[63:38] feeds the infrastructure.
[63:40] It feeds your whole supply chain. It's
[63:42] like just to make it visceral, Greg,
[63:44] it's like imagine you were like, "Cool,
[63:45] I'll pay you a million bucks a year. I
[63:47] want you to do nothing but think deeply
[63:49] about the future of AI and technology
[63:51] and the implications it's going to have
[63:52] on product building. And I just want you
[63:54] to build like the a worldclass long-
[63:56] form content library that all of our
[63:58] different departments can pull from. The
[64:00] robots do the repeating. We can pluck
[64:02] short form. We could turn it into
[64:04] automated videos. Like all the
[64:05] infrastructure, I just want you to
[64:07] create the highest quality input. Now
[64:10] imagine if Salesforce did that with like
[64:13] a Pulitzer Prizewinning journalist or
[64:16] Slack did that. I think that the
[64:18] leverage you get out of that is so it's
[64:21] so hard for people to wrap their heads
[64:23] around and they're so focused on the
[64:24] minutia of like we got I post five times
[64:27] a day and what time of day and how do we
[64:28] go viral and can we pay for retweets and
[64:30] like all none of it matters. None of it
[64:33] matters. What matters is you have the
[64:35] infrastructure and you have a very
[64:36] valuable content library that you pull
[64:38] from and leverage for a long period of
[64:40] time.
[64:41] >> Yeah. I mean that's why people are
[64:42] starting to say the company brain is the
[64:45] moat.
[64:47] >> Yes.
[64:47] >> Because right because if you're uh if
[64:51] you're com like if I hired you Cole and
[64:53] you're adding to this company brain and
[64:55] it's getting more and more valuable and
[64:56] I can use robots to go and slice and
[64:59] dice it a million ways. Like I I know it
[65:02] sounds crazy to pay someone a million
[65:03] dollars, but uh it's not that crazy if
[65:07] the value from that million dollars is
[65:10] thousands of pieces of content. It's
[65:13] improving workflows. It's helping with
[65:15] the product. It's it's pretty insane,
[65:17] right? And that's why you're starting to
[65:18] see actually, you know, companies, you
[65:20] know, spend millions of dollars, tens of
[65:23] millions of dollars some and I think it
[65:25] will be hundreds of millions of dollars
[65:26] on data licensing or buying data. I saw
[65:30] Micro One buying data uh Spirit Airlines
[65:34] data for mill I think millions of
[65:35] dollars
[65:36] >> and you know why because they're going
[65:38] to use that data. They're going to mine
[65:39] that data you know and um I think that's
[65:44] where everything is going.
[65:46] >> That's exactly that's exactly it. It's
[65:48] just people understand this idea more in
[65:50] technology. They go, "Oh, we acquired a
[65:53] a company that just filed for bankruptcy
[65:56] and we bought all their data." And
[65:57] everyone's like, "Yep, I instantly
[65:59] understand the value." The reality is
[66:02] language and writing and libraries of
[66:04] content are data. And you are buying you
[66:08] are buying groupings of words. You are
[66:09] buying ideas. You are buying
[66:11] crystallized thinking that you can
[66:13] leverage the same way you leverage data.
[66:16] >> Yeah. It's funny cuz like I bet the
[66:18] Spear Airlines people are probably like,
[66:20] "Haha, I can't believe they bought our
[66:22] data for millions of dollars, idiots."
[66:26] And the micro one people are like,
[66:28] "Haha, I can't believe they sold us the
[66:30] data for so cheap." [laughter]
[66:32] >> Yeah. Well, it's it's always a question
[66:34] of it's actually not really the the
[66:37] value of those assets in and of
[66:40] themselves. The question is how do you
[66:42] leverage it?
[66:44] >> You know, like if you're if you're a
[66:45] company and you buy some writer's entire
[66:49] catalog and library of content and then
[66:51] you don't have the operational
[66:52] infrastructure to scale it and squeeze
[66:54] all the juice out of it, then that was a
[66:56] horrible decision because you you're not
[66:58] able to do what you should do with it.
[67:00] But if you have a really smart person as
[67:02] your input for your entire content
[67:05] marketing machine, like literally one
[67:08] person doing that for Slack or
[67:11] Salesforce or Airbnb or whoever could
[67:14] potent one smart person
[67:16] >> could build the entire digital brain
[67:19] library of content that their entire
[67:22] marketing machine uses and re-wizzles
[67:24] and remixes and posts and chops and cuts
[67:26] and paste. Like that's to me that's the
[67:29] future.
[67:30] >> I mean, we're out of time, but that's
[67:31] like a whole other startup idea that
[67:34] someone needs to do, which is just do
[67:36] company brains as a service.
[67:39] >> Come on.
[67:39] >> Yep.
[67:41] >> It's a it's it's a home run. If I if I
[67:43] was excited at all about B2B, I would go
[67:45] do it, but I'm not.
[67:47] >> Cole, you've been so generous with your
[67:49] ideas, the whole playbook. You you open
[67:52] sourced it. I really appreciate it.
[67:54] Where could people get to know you
[67:56] better, find you, and get more of your
[67:58] ideas?
[68:00] >> Always always pumped to chat, my man.
[68:02] Um, you can find me on X. I still want
[68:04] to call it Twitter. You know, I'm a I'm
[68:07] an old head. Um, you can find me on X.
[68:09] And if anyone wants to build their
[68:11] digital brain and library of content,
[68:13] you should come join us for the next
[68:14] Ship 30 for30 cohort. We're going to do
[68:16] that in October. Should be a lot of fun.
[68:18] >> Cool. Yeah, I'll include links uh for
[68:20] all that in the show notes in the
[68:22] description. Cole, you've been on the
[68:24] pod probably what, three, four times?
[68:27] >> Is it three? Three or four? Yeah.
[68:29] >> I love having you on. I hope this
[68:31] episode is well well loved. Um it's one
[68:35] of those things where this is like you
[68:37] were talking about timeliness, timely
[68:39] versus timeless. And like this is a
[68:41] timeless episode. like people who ingest
[68:44] this these ideas 2026 now like 2029 2032
[68:49] like these are these are ideas that are
[68:51] always going to to be so so valuable and
[68:56] it's it's sort of in some ways like an
[68:59] eat your vegetables episode
[69:02] [laughter]
[69:02] >> you know versus like drinking a smoothie
[69:05] it there's just so much nutrients in
[69:07] this and I do believe that people who
[69:09] actually you know if you've made it
[69:11] here. The people actually who who've
[69:13] made it here uh like they are going to
[69:16] understand how to have strong points of
[69:18] view, how to use AI to create some of
[69:19] this content, how to think about how it
[69:21] all plays together. And uh they're
[69:24] lucky.
[69:26] >> Well, thanks, man. Yeah, that's my goal.
[69:28] As timeless as possible. [sighs]
[69:30] >> I'll catch you later, Cole. Thanks.
[69:32] >> Thanks, everyone.