YT YouTube2Text

YouTube2Text → Transcripts → $30M Writer: Never write AI slop again

$30M Writer: Never write AI slop again — Transcript

by Greg Isenberg · 11,398 words · 2,055 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Nicholas Coles made over 30 million
0:02dollars writing on the internet. So when
0:04he talks about
0:05content, I listen. And this episode
0:07honestly felt less like an interview and
0:10more like just a free
0:11masterclass. He breaks down why most
0:14people's content gets ignored in the AI
0:17era, why AI is
0:18making generic writing worthless, and how
0:20to build what he calls a digital brain. A
0:23library of your
0:24stories and your opinions and your
0:26lessons and ideas that gets you customers
0:2924-7, that builds
0:30you trust and gives AI something real to
0:32work with. If you write online, you sell
0:35online, or you want
0:36ideas to compound, this is honestly worth
0:39thousands of dollars, but it's free for
0:41you. The people that
0:42stick around to the end of this are not
0:45going to be writing AI slop. And I think
0:48the people that
0:49you know, use AI to actually create
0:51interesting content. I mean, of course,
0:54those are the people
0:55that have the highest probability of
0:57success. Enjoy the episode. I learned a
0:59lot. I think you
1:00will too. Today's episode is brought to
1:02you by Brex. My company's been on Brex
1:04for
1:04a year and a
1:05half. And I started because I kept
1:06hearing companies like Vercel, OpenAI,
1:09Anthropic were
1:09using Brex. And I figured if they're
1:12using it, why shouldn't I? It's been a
1:14game changer. The
1:15thing that got me is how smooth it is.
1:17It's got high limit cards. It's got
1:19banking. It's got AI
1:20that handles the back office busy work
1:22like expense reports, which I don't want
1:25to do on its own.
1:26It's really just built for this agentic
1:28world. If you're building something new,
1:30it's time to get
1:31Brex. Check it out at brex.com slash
1:33solutions slash startups. Link in the
1:35description.
1:45Nicholas Cole, thank God you're here,
1:47man.
1:48By the end of this episode, what are
1:50people going to learn?
1:52By the end of this episode, you will
1:54better understand the short, medium, and
1:56long-term
1:57value of everything you create and write
2:00and market on the internet before you
2:03create it.
2:04That is the goal.
2:04Before you invest the time, you will
2:06understand why you're doing what you're
2:08doing in the
2:08first place.
2:10Yeah, and I think that matters a lot
2:12because all of us are vibe coding
2:14software.
2:14We know there's a lot of opportunity to
2:17build apps, ship SaaS, things like that.
2:20But it's hard to get customers.
2:22And what I love about you is through
2:24writing, you understand how to get seen,
2:27how to create stuff that goes viral.
2:30And there's sort of a bit of an art to
2:32it.
2:33And what I hope to learn, selfishly, by
2:37the end of this episode, is to get some
2:40of
2:40that art from you.
2:42to learn about that so that the products
2:45that I create, I can accompany it with
2:47content writing
2:49that tells stories that hopefully goes
2:52viral or that just resonates with people.
2:56And Cole, can you commit that to me? Will
2:59I learn that?
3:01Yeah, that's what we're going to talk
3:02about.
3:02So something I see people talk about a
3:06lot these days is that in the AI era,
3:09SaaS no longer has a moat. And I actually
3:13think that that's not true. I think that
3:17SaaS has a moat when you have a point of
3:19view. And I didn't, that was a nice
3:22little
3:23rhyme there.
3:24But what I mean by that is if you're just
3:26building commodity SaaS, meaning like
3:28you're just providing
3:29features that accomplish tasks to people,
3:33that's fundamentally a commodity.
3:35SaaS with a point of view means that
3:39you're building some sort of product or
3:42platform that enables users to do things
3:46in a certain way.
3:48Like they're executing a certain method.
3:51So, for example, like we have a SaaS
3:53platform called TypeShare.
3:56And the goal of TypeShare is not, well,
3:58there's a gazillion scheduling apps and
4:00there's a gazillion writing apps and our
4:02writing app does writing and scheduling
4:04too.
4:05Right. Like that is a commodity. The goal
4:08of TypeShare is to, quote unquote,
4:11automate or or make tasks easier for
4:13people who want to execute a different
4:17methodology for writing on the Internet.
4:21So there's a point of view that drives
4:23the product. And if someone says, well,
4:25why would that create a moat?
4:27for example if we say inside type share
4:29well your goal is we're going to help you
4:32build your personal
4:33language model and our point of view is
4:36that there are certain types of content
4:38commodity content
4:39personality content original content each
4:41of these different types of content
4:42create
4:43different
4:43leverage inside of your personal language
4:45model even as i'm articulating that
4:47imagine you're trying
4:48to build a sass platform that does
4:50everything our sass platform does every
4:53time you build a feature
4:54It's like you have to turn back and look
4:56at me and go, oh, but wait, what does
4:58that
4:58actually mean?
5:00What's the point of a personal language
5:01model?
5:02What does commodity content mean?
5:04So you will forever be one step, two
5:06steps, three steps, five steps, ten steps
5:08behind because you constantly have to
5:10look
5:10at the person who's creating the point of
5:12view or creating the methodology to go,
5:15wait, wait, wait.
5:16They're the one doing the thinking.
5:18I'm just going to copy them, which means
5:19you're always behind.
5:21And so I think it's funny when people say
5:22there's no moat in SaaS because it's
5:24like,
5:24well, yeah, there's no moat in anything
5:26if
5:26you don't have a point of view.
5:27But there's a ton of moat in SaaS and
5:29product and writing and marketing if you
5:32do have a point of view.
5:33And so that's kind of what inspired this
5:36convo is like, so how do you create a
5:37point of view in the first place?
5:40So where do we start?
5:41So an idea I've been thinking about a lot
5:45is the fact that the English language or
5:49any alphabet, any language, is really
5:53just
5:53an open source language for
5:56communication.
5:58Meaning no one owns the individual
6:01letters of the alphabet.
6:03No one owns words, like the vast majority
6:06of words, right?
6:08Like you don't own the word tomato.
6:11I don't own the word catastrophe.
6:14We also don't own even our names.
6:18There are a lot of people in the world
6:19named Greg.
6:20There are a lot of people in the world
6:21named Cole.
6:22And if someone says, my name is Cole, or
6:25my name is Greg, and you write, my name
6:28is
6:28Greg,
6:29right, you're not plagiarizing them, and
6:30they're not plagiarizing you.
6:32And as you start to, like, tease that
6:35idea out, you realize that the vast
6:38majority of
6:39words and groupings of words and thoughts
6:42and ideas and sentences are shared. They
6:46are not
6:46actually attributable to any one
6:48individual. And I think this is what
6:51people are wrestling with
6:52when they read writing that they think
6:55was written by AI, is when you identify,
6:59quote-unquote,
7:00AI slop, what you're really identifying
7:03is commodity content, meaning you're
7:05reading something,
7:07but it feels the way that it does because
7:09it's not attributable to any one
7:10individual.
7:12It's like reading health and wellness
7:13content.
7:13Someone writes, the key to losing weight
7:16is eating healthy.
7:18That's a commodity sentence.
7:20There's a gazillion people who have said
7:22that exact grouping of words.
7:24And we don't look at the millions and
7:26millions of people who have repeated that
7:28same grouping
7:28of words and go, oh, everyone's
7:30plagiarizing each other.
7:31We just go, that's a commodity sentence.
7:33and so as you like start to tease this
7:36out it raises a really interesting
7:38question in the age
7:39of ai and the age of ai is the creation
7:42of infinite text-based content anyone can
7:45write
7:46anything on any topic at any time with
7:48any amount of volume and speed they want
7:50which raises the question so what is
7:53worth writing in the first place
7:55and i think it's it's interesting even
7:58like your content is a really good
8:00example
8:01of this where
8:02the simplest delineation is when you're
8:05writing and you say the difference
8:07between
8:08I think,
8:09I did this, I perspective versus you
8:12should. You should do this. Because when
8:16you're talking to the
8:18reader through the lens of you,
8:20inherently the content is going to lean
8:23more commodity.
8:25Because you're kind of just like
8:27repeating all the things that everybody
8:29knows everyone should do.
8:30But when you're writing from a place of
8:32I, like I just did this or I think this
8:34or
8:35you share all the time, it's like I think
8:37this would be a great startup idea.
8:40When someone consumes that content, that
8:43is personality content.
8:45That is I am consuming what Greg thinks
8:48about the world, not what everyone in
8:51this
8:51topic area typically says about this
8:54topic
8:55in the world.
8:57You know what I mean?
8:57Yeah, I mean, that's why all content
8:59isn't created equal, because sometimes
9:02I'll see people go insanely viral on X,
9:05let's say, and I look at the content and
9:07I'm like, I've read this before. It's a
9:09Wikipedia article essentially summarized,
9:12or it's just a take that's been said
9:15before, and they plopped it into ChatGVT
9:17and asked to create an essay on it.
9:19and it could still get 5,000 likes or
9:2310,000 likes,
9:25but at the end of the day,
9:27it's like, what game are you trying to
9:28play?
9:28When I'm creating content, I think about
9:31two things.
9:32I think about how do I drive as much
9:36value
9:37for the person consuming the content?
9:40So I'm just constantly, as I'm writing,
9:42I'm thinking about the person in the
9:44other seat
9:45and thinking about how can I help them?
9:48and that frame of mind is super helpful
9:51because that means actually you're not
9:53going to write things like
9:54you should do XYZ, you should do XYZ
9:57because that's kind of coming from a
9:59place of like
10:00I'm better than you and I know everything
10:04therefore you should do this and I don't
10:05think that's what I try
10:06that's what I want to avoid
10:08and then the second thing is creating
10:11affinity
10:11so
10:13So, you know, if I add value to people's
10:16lives, it's sort of an effective number
10:19one.
10:20It generates some affinity to, hey, this
10:23guy helped me out a little bit.
10:25You know, I got an idea from him or he,
10:28you know, I never I never thought in, you
10:30know, of this topic in that way.
10:33Like, for example, what you were talking
10:34about with like the English language is
10:36open source.
10:37I never, you know, I use English every
10:39day of my life.
10:40I never thought about that concept
10:42before.
10:43And for the rest of the time, I'm going
10:45to be associating that with you, Cole.
10:48And I think people undervalue how
10:52valuable it is to have a new idea and
10:57ship
10:57that value to someone.
11:00And it actually affects their day-to-day,
11:02especially within the context of a
11:04business context,
11:05because that means that they're going to
11:06be making more money.
11:07They're going to be being more
11:08productive.
11:09and that's going to put food on their
11:12plates,
11:13feed their family, get them a job
11:14promotion.
11:16There's so much good to that.
11:18And that's what I want to unpack today
11:19is that piece for people.
11:23Yeah, so, okay, a ton of thoughts.
11:26First, value is always relative to where
11:31the audience is,
11:33meaning something that you could find
11:36really valuable
11:37because maybe you're a more advanced
11:38entrepreneur,
11:40a beginner would not find as valuable
11:42because they're not far enough along yet.
11:44And in reverse, something that's very
11:47basic to you
11:48that you've known and you've learned that
11:50lesson a thousand times
11:51is not valuable to you anymore because
11:53you already know it.
11:54But to a beginner, that's like the most
11:56helpful thing
11:57you could give them is step one.
11:59So value is always relative to where the
12:02other person is.
12:03what prompted this whole mode of thinking
12:06for me is, especially in the past two or
12:09three years,
12:10and as I've seen AI start to accelerate,
12:13I've found myself living in a constant
12:15state of
12:16frustration where I know that these
12:19models can write a lot of things just as
12:22well as I can.
12:23My frustration is knowing that that
12:27solution exists and yet I still find
12:30myself spending time writing and
12:33rewriting
12:34things to tell beginners that I know
12:37would
12:38be helpful to them but are really boring
12:41for me to repeat. And so I've been
12:44writing nonfiction online for 15 years.
12:47I've been
12:48writing about how to write on the
12:50internet for over a decade. I'm not
12:53exaggerating,
12:54like probably written, I don't know,
12:57north of 50 million words on the topic,
13:00like thousands and
13:01thousands and thousands of pieces. And
13:03especially in the past few years, this
13:05frustration for me is
13:07like, I know that there are still a
13:09tremendous amount of people in the world
13:12who need help
13:12taking the first step and going, how do I
13:15start writing on the internet? What's my
13:17niche? How do
13:18I name and claim what makes me different?
13:21Ideas that are very simple to me that are
13:24very novel to
13:25beginners. And yet every time I write
13:28about them for the 999th time, I feel
13:31like
13:32I'm wasting my
13:33finite time on earth because I'm just
13:36re-swizzling the same groupings of words
13:38over and over and over
13:40again. And so what kind of that, that's
13:42what originally prompted this way of
13:44thinking.
13:45and it made me realize that commodity
13:48content
13:49doesn't have zero value.
13:51Commodity content does have value.
13:54And for a lot of people,
13:55particularly beginners in any industry,
13:58commodity content is actually the most
13:59helpful thing
14:00you could give them.
14:01The problem is that as you write more and
14:04more commodity content
14:06about a specific topic or in a niche,
14:09you will feel what I'm feeling.
14:11You probably already do, where you're
14:12like,
14:12I find myself just repeating the same
14:14ideas that I've already thought through.
14:16I'm not actually thinking through
14:18anything new.
14:18I'm just repeating things I already know.
14:22And that is very low leverage.
14:25And so as you begin to build your content
14:29library, you have to sort of examine why
14:32am
14:33I writing the things that I'm writing?
14:34And each new thing that I write, what is
14:37the additive value of the thing that I
14:40just created to my what I'll call digital
14:44brain, my personal language model?
14:48And in an ideal world, like so just a
14:51quick little backstory.
14:53I've been ghostwriting for a decade.
14:55I used to do this framework with clients
14:58where when I would be working with a
15:00client who's really, really busy, maybe
15:03they wrote a book.
15:04Maybe they would have a podcast.
15:06More likely, maybe they would just record
15:08internal meetings.
15:10And I would pull from those assets, pull
15:12from their book or pull from blog posts
15:14or
15:14pull
15:14from things that they had created or said
15:16out loud.
15:17And I would take those words and phrases
15:19and I would assemble them into content.
15:22And that's essentially what a ghostwriter
15:23does.
15:24The misunderstanding is that the
15:26ghostwriter isn't really creating new
15:29ideas.
15:29The ghostwriter is not really like
15:31putting net new thinking in the client's
15:33mouth.
15:33the ghostwriter is essentially taking
15:36what I like to call approved language.
15:39The client said it.
15:41And so why can't I take a sentence from
15:43chapter one and a sentence from chapter
15:46two
15:46and two sentences from chapter three and
15:48two sentences from chapter four of their
15:50book
15:51and stitch them all together and create
15:52something new?
15:54The client said all of those words in
15:56those groupings of words.
15:58And so I did the manual version of what
16:01AI can and should do here soon, which is
16:04the larger your library grows, the more
16:08ideas and approved language you have, the
16:11more net new combinations you can create.
16:15And I think that that fundamental idea is
16:17something that the masses do not
16:19understand about the leverage you get in
16:22crystallizing your thinking and building
16:25a
16:25library of content on the Internet,
16:27because the AI is not actually writing.
16:30AI is taking approved language from
16:32things that you've written and creating
16:35net new combinations.
16:37And if you play that out, the benefit of
16:40doing that is now I can continue to help
16:43beginners,
16:44telling them the ideas I know they need
16:46to hear, but I don't need to take my
16:48finite time
16:49on earth rewriting the same thing over
16:52and over again.
16:53Which, what does that mean?
16:55That opens up more time for me to do
16:57what?
16:57to think about the next thing I haven't
16:59thought about.
17:01And I think this is the hope,
17:03or this is the utopian hope of AI,
17:06is that AI will automate the low leverage
17:09stuff,
17:10which opens up time for higher leverage
17:12things.
17:13And so in the case of writing,
17:15what that means is AI will help me repeat
17:17the same commodity ideas
17:18that I've already written 10,000 times,
17:21which opens up time for me to think about
17:23non-commodity ideas, personality ideas,
17:26or ideally original ideas,
17:28which is more additive to the long-term
17:30value
17:30of my digital brain,
17:32which imagine the societal implications
17:34of more people thinking about things
17:37they haven't thought about before
17:39rather than just like churning out blog
17:40posts
17:41of things that they've already thought
17:42about
17:42and just need to repeat.
17:44To me, that creates two bottlenecks
17:47and I'm curious your thoughts on this.
17:49The first is how do you create a second
17:53brain
17:54that people use Obsidian or whatever
17:57second brain software you want to use
17:59so that you're properly capturing those
18:03ideas.
18:04Those ideas are in a safe place.
18:09And the second bottleneck is, okay, I've
18:12got these high-quality ideas.
18:14Actually, some of them might be
18:15high-quality, some might be low-quality.
18:17But I've got these ideas in this bank.
18:20How do I essentially plug it into an LLM
18:24such that I don't get slothed?
18:27There's a bunch of people who are just
18:32able to mangle their LLMs such that they
18:37understand how to prompt it in a way or
18:42have agents run in a way such that it's
18:46eloquent.
18:48So I think to me, those are the two
18:50bottlenecks.
18:50Am I missing something?
18:52Yeah, I mean, I think that there's
18:54certainly outlier use cases.
18:56And I'm not against the sort of person
18:58who is not a writer, is not a proficient
19:01writer,
19:02doesn't want to be a proficient writer,
19:04but knows a lot about an industry,
19:05wants to brain dump into an LLM and then
19:08have the LLM take that information
19:11and just write it the way that the LLM
19:13would.
19:14That's really no different than how
19:16ghostwriters work with clients.
19:18And inherently, in hiring a ghostwriter,
19:20you get subjectivity, right? Ghostwriters
19:22are human beings.
19:23They all write in different ways. You're
19:24going to get a subjective output because
19:26you're working
19:27with a specific ghostwriter. An LLM is no
19:29different. And I'm not against that way
19:31of
19:31working.
19:32I think that the fundamental mistake and
19:35the point of view that I'm really leaning
19:37into
19:37for the foreseeable future is that the
19:40highest leverage thing you can do is the
19:43writing.
19:45The robots should do everything else.
19:48And I think where people get lost and why
19:51slop happens is they might agree or
19:54understand
19:56that idea, but then they want to skip the
19:59step of writing and go, well, my LLM
20:02knows
20:02enough and it can just do it for me.
20:04And you don't realize that in doing that,
20:06you're giving up a lot of leverage.
20:08I don't want the robots writing for me.
20:11What I want is the robots repeating me,
20:14which is different.
20:16What does that mean?
20:16So, for example, you post on a bunch of
20:19different social platforms, right?
20:23And I know that I'm like, okay, I have
20:26this giant library of short-form text.
20:29Giant.
20:30I've tweeted 10,000 times, whatever it
20:32is, right?
20:33And I want to keep publishing two
20:35short-form pieces a day, and I also want
20:37to put out short-form on Instagram
20:39threads
20:40and YouTube community posts and some of
20:41these other platforms.
20:43It is extremely low leverage for me to go
20:47through my library
20:48and find the highest performing posts and
20:50repost them on multiple platforms.
20:53People got around this with a social
20:55media manager.
20:57Cheaper solution, you get around that
20:59with a VA.
20:59And even cheaper solution, now AI and
21:01agents can essentially do that.
21:04But the step further and where I see
21:06long-term value
21:07in this idea of building a digital brain
21:10or a personal language model is that it
21:13is
21:13not
21:13difficult for AI to understand.
21:16Cole, you wrote this short form piece
21:20about five benefits of journaling, and
21:24you
21:24wrote
21:25this other short form post on seven
21:27benefits of journaling.
21:30And I can take some from the first list
21:33and some from the second list and create
21:35a
21:35new
21:35third list that is just repeating a
21:37different combination of those two posts.
21:41And this is where every time you sit down
21:43to write, I think one of the most helpful
21:45things
21:45to ask yourself is, what is the short,
21:47medium, and long-term value of the thing
21:49that I'm creating?
21:51And I'll give you an example of, so what
21:54is the value of you writing commodity
21:56content?
21:57If commodity content is a commodity,
22:01what's the point? The point is that there
22:04is
22:04borderline infinite commodity content
22:07inside every topic. The question is not,
22:10how do you write about every type of
22:13commodity content in your topic? The
22:16question is,
22:17which types of commodity content do you
22:19agree with? And which types do you
22:21disagree with?
22:22So I'll give you a simple example. In the
22:24world of personal finance, two pieces of
22:27commodity content
22:28are if you want to build wealth,
22:31you should skip your morning coffee.
22:34Don't waste that $3.
22:35And you should just buy the S&P 500
22:37and 30 years from now, you'll be a
22:38millionaire.
22:39Can I guess a second?
22:41Yeah, yeah.
22:42Because I'm just curious.
22:44Is the other one,
22:45if you only rented and didn't pay a
22:47mortgage,
22:49you would actually outperform in general?
22:51I feel like I see that one all the time.
22:55Yeah, so these are examples
22:57of commodity ideas.
22:59These are not unique to any one
23:00individual, right?
23:02Like one person says, save your coffee,
23:04buy the S&P, you'll get rich.
23:06Another person says, saving your coffee
23:07money
23:08isn't gonna make you rich, you're focused
23:09on the wrong thing.
23:10Both of those are commodity ideas.
23:13The question and why you should write
23:15about it
23:15is which one do you agree with?
23:18Because those are two different ways of
23:20looking at the world.
23:22And so where people run into, oh, AI is
23:25creating slop for me
23:26is they don't realize how many of those
23:29micro decisions they're expecting AI to
23:31just understand
23:33about them. So AI might like, let's say
23:35their topic is personal finance. AI might
23:37write something
23:38and use the example, well, you should
23:41save your $3 of coffee money and just
23:44invest in the S&P 500.
23:46And the person reads that and goes, oh,
23:48this is slop. But they don't realize that
23:50the reason AI
23:51wrote that is because AI doesn't know
23:53which perspective you agree with. So what
23:56is the
23:57long-term value of your library? Well, on
23:59the commodity content side, the long-term
24:01value is
24:02you actually yourself having clarity over
24:04which points of view do you agree with,
24:06which points of
24:07view do you disagree with. It's not the
24:09ideas themselves that are unique. It's
24:10which ones are
24:11tied to your identity. And the more of
24:13those that you crystallize, the more
24:15long-term leverage you
24:17get out of working with AI because AI
24:18knows, oh, I'm never going to use that
24:20example because
24:21that's not how you see the world. How do
24:24you actually set this up? If someone
24:27listening to
24:27this is like, I agree, I agree with Cole
24:30and what he's saying, it makes a lot of
24:32sense.
24:34How do I actually set this up? Well,
24:37shameless plug, this is the point of
24:40view.
24:40This is what
24:41we're building inside of TypeShare. So if
24:43you want SaaS that does that, there you
24:45go.
24:45But the reality is there's a lot of
24:47different ways that you can accomplish
24:49this.
24:50And fundamentally, it actually doesn't
24:51really matter where the library of
24:53content
24:53lives.
24:54What matters is that AI has access to it
24:57And more importantly you can skip the
25:00step
25:00of understanding what you doing in the
25:03first place I fundamentally believe that
25:06the reason people are struggling with
25:09writing an AI slop and writing with AI is
25:12because deep down everyone wants to skip
25:15the step of thinking and just have the
25:18robots do the thinking for them
25:20and in doing so it's like you're trying
25:22to find a shortcut and the shortcut is
25:24the
25:25longer road in
25:25disguise and so like an example that's
25:28very visceral for me is when i was
25:31writing
25:32on quora
25:32back in 2013 2014 i wrote a handful of
25:36answers about me playing world of
25:39warcraft
25:39professionally
25:41as a teenager and i would always like in
25:43each answer i sort of referenced it in
25:45different ways
25:46I was like, oh, when I was a teenager, I
25:48was a really competitive World of
25:49Warcraft
25:50player.
25:50And then another piece, I would say the
25:51same thing, just like a different
25:53combination of words.
25:54And one day I wrote a piece and the first
25:56sentence was, when I was 17 years old, I
25:59was one of the highest ranked World of
26:01Warcraft players in North America.
26:04And that Quora answer went viral.
26:06And so like me, myself, manually looked
26:09at that and when that first sentence and
26:13that
26:14way, that grouping of words, that way of
26:16talking about that experience, my
26:19hypothesis is clearly
26:20that resonated with people.
26:22So what did I do?
26:24I wrote hundreds of other pieces over the
26:28next five, 10 years, repeating that exact
26:32sentence somewhere.
26:33Maybe it was the first sentence. Maybe it
26:36was a hook somewhere in it. Maybe it was
26:39before I told a story. But I used that
26:41grouping of words hundreds of times. That
26:44is the goal of what AI can and should
26:46help
26:46you leverage, is that as you get clarity
26:49over these ideas and groupings of words
26:51and points of view and perspectives, once
26:54you say it really well once or twice or
26:57two different variations and you like
26:59both, AI can then go, yeah, you've said
27:02that. I got it from here.
27:03You want AI doing repeating.
27:06You don't want AI doing net new writing,
27:09thinking,
27:10I'm assembling groupings of words and
27:12groupings of sentences
27:13that you've never done before.
27:15That's why then it gives you something
27:17and you're like,
27:18this isn't me.
27:20Because it's not you.
27:23Yeah, and people sniff it out too.
27:26And it actually hurts you more than it
27:29helps you.
27:29because then they associate you with, in
27:33my opinion,
27:35then they associate you with someone who
27:37isn't adding value.
27:38And it's like, there's nothing worse.
27:42I don't think there's anything worse that
27:44you can do
27:45than steal someone's time.
27:46And I think that's why people are so, you
27:49know,
27:49hate slop so much.
27:50It's because they consume this piece of
27:52content
27:53and they're like, wow, that was sloppy.
27:55I hated that.
27:57Mm hmm. I think this is a second really
28:00fascinating question in the age of AI,
28:04which is so what is ownership if if
28:07language is open source?
28:10No one owns the alphabet. No one owns
28:11ninety nine point nine percent of the
28:13words.
28:14And most people don't even own ninety
28:16nine point nine percent of the groupings
28:18of words.
28:19If language is open source, what is
28:22ownership and ownership?
28:25and obviously there's like a legal
28:27component.
28:28It's like you can trademark names
28:30and like there is a legal component.
28:32But generally speaking,
28:34ownership is really just a different word
28:37for association.
28:39You associate ideas with certain people.
28:44Now you can achieve association in lots
28:46of different ways.
28:47The easiest way to achieve association
28:49is to write about a topic a lot.
28:52like people associate ai startup ideas
28:56with you and we can just say at a
28:59fundamental level for no
29:02other reason than you create more about
29:04that topic than the vast majority of
29:06people or i'll give you
29:07i'll give you another example two
29:10examples of your coffee drinking coffee
29:12and wasting money which
29:14is when I hear that concept, I think of
29:18Ramit Sethi, I will teach you to be rich,
29:22who always
29:23says, keep drinking your coffee, even if
29:26it costs $6, live your rich life. And
29:29that's against the
29:31common idea of save, put in the S&P. And
29:34the other side of that coin is Dave
29:37Ramsey, who's like,
29:38save, invest, don't go in debt. And so
29:42you, you know, both ends of the spectrum,
29:45same core idea.
29:48Exactly. And I mean, so to build on that,
29:50right, Dave Ramsey is known for that
29:53meme,
29:54where he's like taking the live call. And
29:56he's like, so you spent all your money on
29:58what? And
29:58then it's a funny answer. And he's like,
30:00and you owe how much money to who? And
30:02then it's a funny
30:03answer. Like what's actually happening is
30:06that that's a format, that's a structure
30:09that people
30:10associate with him. So ownership is
30:14really just association, generally
30:18speaking. And the easiest
30:20way to establish association is through
30:23volume. You write a lot about a topic,
30:26people associate
30:27that topic with you. Very basic. The next
30:31layer up of association is personality.
30:36It
30:36is, okay, so
30:37it's not just that you write about this
30:40topic, but the audience knows all of
30:42these
30:42specific
30:43things about you. They know, like, so
30:46play it out. It's like you have a, like a
30:50personal finance
30:52expert and you know nothing about them.
30:53They just write a lot about personal
30:55finance. And then
30:56there's another personal finance expert
30:58who is maybe just as qualified or less
31:00qualified,
31:01but they share a lot about their personal
31:03life, their personality, how they see the
31:05world.
31:06She's a personal finance expert. She's a
31:08single mom. She got divorced three years
31:10ago. She has
31:11two kids. She's had to work multiple
31:13jobs, put them both through school.
31:15Really
31:16hard worker,
31:17started a blog, blog took off. You know
31:20all these things about that person.
31:22And something that I think a lot of
31:26people forget is that, quote unquote,
31:29less
31:30credible, more personality content
31:34typically outperforms more credible
31:37commodity content.
31:40So like, take the extreme, like there are
31:43not as successful personal finance people
31:46or finance people with significant
31:49personality content that get more
31:52distribution and have bigger audiences
31:55than Ray Dalio.
31:57It was like a multi-billionaire, right?
32:00So the credibility of the person is
32:02actually not like, it's important.
32:04You want to learn or hear from someone
32:06who has done something and knows about
32:08the
32:09thing.
32:10But I think people forget that
32:12credibility itself is not the winning
32:14card.
32:15The winning card is actually the
32:17personality of the person and how much
32:19they're willing
32:20to let you into their world and how many
32:22things about them you associate with that
32:25topic.
32:26And that idea should be really liberating
32:28for a lot of people.
32:30Like I'm sure a lot of your listeners,
32:31people that build products or are
32:33first-time founders
32:33or second-time founders, they want to
32:35build organic audiences.
32:36like you actually don't need to be coming
32:39up with game changing, radically new
32:41original thinking.
32:44All you need to do is just go, yeah,
32:45there's a gazillion startup founders, but
32:48I am a startup
32:48founder. And these are this is what makes
32:50me unique. This is how I see the world.
32:52This is
32:52where I live. This is how many siblings I
32:54have. This is where I went to school.
32:56This
32:56is what I
32:56studied. This is what my hobbies are. And
32:59all of those little data points actually
33:01now in the age
33:02of infinite text-based commodity content,
33:04all of those little data points are
33:06differentiators.
33:08And more importantly, every time you
33:09write about one of those personality
33:11details,
33:13that personality detail should make its
33:15way into your digital brain, your
33:16personal
33:17language model.
33:18Just like with me, when I was 17 years
33:20old, I was one of the highest ranked
33:21World
33:21of Warcraft
33:22players in North America. There's a
33:24reason why I repeated that 500 times, you
33:27know? And so then
33:28association is really just I associate
33:31these life, life details, life events,
33:34ways of seeing the world
33:35with you. Like, I'm sure you're familiar
33:38with Gary V repeated this has been
33:41repeating this for 15
33:42years, which is I want to buy the New
33:45York Jets over and over and over and over
33:48and over again.
33:49Like, what is that? That's a personality
33:51detail. And you associate that detail
33:52with
33:52him.
33:53And I think he lives in New Jersey.
33:57So it's kind of funny, right?
33:59I know he works, maybe he has a place in
34:01Manhattan,
34:02but I think he grew up and still lives in
34:05New Jersey.
34:06So it is interesting that you can have
34:08someone
34:08who lives in another place
34:10and his dream is to buy the New York Jets
34:13and tens of millions of people associate
34:16Gary Vee,
34:17who's not necessarily a sports person.
34:21I know he has a sports business now,
34:23but made his wealth in wine and agencies,
34:26social agencies,
34:28all of a sudden people connect him with
34:31the New York Jets.
34:33I think what's something really
34:35interesting you said
34:36is basically like, hey, your competition,
34:41basically,
34:43is going to just plop ideas to create
34:46commoditized content.
34:49And there's just going to be more and
34:51more of that.
34:51Whereas if you could do an audit of
34:54interesting tidbits about you, your
34:57brand,
34:57how you think about things, point of
35:00views
35:01as you call it, throw it in a markdown
35:03file or use a second brain software like
35:07Obsidian or something.
35:08and it's going to just stand out
35:14by virtue of everyone else
35:16is using the same tool
35:19and the same methodology to crank out
35:22content.
35:23So it's like, it's just,
35:25it's sort of like fish where the fish are
35:28and you're going to be able to fish with
35:32dynamite
35:33if you actually have these stories and
35:34stuff like that.
35:35Yeah, to sharpen the precision of the
35:38language, it's that everyone is using the
35:42same LLMs, meaning they are all creating
35:46from the same data set.
35:48Yeah.
35:48Your personality content, your library of
35:52content is the unmade data set.
35:56You can't expect AI to write a paragraph
35:58inside of an essay or a newsletter or
36:01something that you're writing about what
36:04you learned as a kid going through an
36:06experience that you've never articulated.
36:10AI doesn't know that.
36:11That is an unmade data set.
36:13And so the way I think about personality
36:15content and that aspect of a personal
36:18language model is that you want to spend
36:20your time building the unmade data set.
36:23And the way that you do that is by
36:25writing about how you became you.
36:27Like, what are your formative
36:28experiences?
36:29What lessons did you learn?
36:30What mistakes did you make?
36:32All of these things seem very mundane to
36:35ourselves, but that's because we lived
36:38it.
36:38We experienced it.
36:39We take it for granted.
36:40But you articulating a visceral mistake,
36:42especially in a story, instantly
36:44differentiates that piece of writing from
36:47everybody else writing about the same
36:49fundamental commodity idea.
36:52Right. And so that's where you want your
36:55time to be spent is you want to you want
36:57to be building these net new additive
36:59data
36:59points, the unmade data set about how you
37:01became you.
37:02What do you say to people who are
37:04thinking to themselves, well, I just, I
37:07lead an uninteresting life and I don't
37:09have anything interesting, you know, or
37:12specific or unique about me?
37:14Yeah I mean A that incorrect And B like I
37:17don know any other way to say it It just
37:20wrong And B it like you have to get
37:23outside of yourself and realize that even
37:27the most mundane experiences are things
37:30that other people resonate with and see
37:34themselves in
37:35Like a really simple example is I moved
37:37across the country twice in the past four
37:39years.
37:41That's not like the most unique and
37:43different and captivating data point
37:46ever,
37:46but it is an experience.
37:48and I bet if I was at a dinner party and
37:50I said, yeah, I moved coast to coast
37:52twice
37:53in the past
37:53four years, I might be talking with
37:55someone else and they're like, oh my
37:57gosh,
37:57I did the same thing
37:58six years ago. Wasn't it crazy? And then
38:00we're telling each other stories and all
38:01of a sudden
38:02we're like friends at a dinner party. We
38:04fail to realize that we do all of these
38:07things in
38:07conversation every single day of our
38:10lives. And the build on idea, and I've
38:14been thinking about
38:15this for a decade because I was trying to
38:18explain this to ghostwriting clients back
38:20in 2016, 2017,
38:21which is in real life, we walk around as
38:25a walking, talking, personal language
38:30model
38:31already. And what happens is we go have a
38:34coffee meeting. I think about the first
38:37time we met up
38:38for coffee in Miami and you go have a
38:40coffee meeting or meet up with someone.
38:43And what do you
38:44do. You start sharing things from your
38:46personal language model. You're like,
38:48here's where I'm from.
38:51Here's how many siblings I have. I moved
38:53to the city two years ago. Here's what
38:55I'm
38:55into. Oh,
38:56you're into the same thing. We're just
38:58doing the unscalable version of that.
39:02Writing on the Internet is the idea that
39:04instead of repeating the same life facts
39:07and stories
39:08and perspectives over and over again
39:10manually at one meeting and then another
39:12dinner party and then
39:14another grouping with friends and then a
39:16picnic one afternoon. Instead of doing
39:17that manually,
39:19what you can do is you can write about it
39:20once and then publish it on the internet
39:22and have that same coffee meeting sharing
39:25that same perspective or insight or life
39:28fact or story
39:28with borderline infinite people. And then
39:31with AI, that piece now gets added into
39:34your library.
39:35And now you can go, I don't want to tell
39:37that story anymore. I've already told
39:39that
39:40story 10
39:40times. I've written about it 10 times.
39:42We've got 10 different variations. I just
39:43want you to
39:44repeat it for me. And that's, that is the
39:47fundamental leverage of writing with AI
39:50in
39:50for,
39:51for the next decade, foreseeable future.
39:53So that works with personal brand. Like
39:55if
39:55you're trying
39:56to build a personal brand, you want to
39:59get out there, you want to be Gary V that
40:01works. How does
40:02this work with branded content? First of
40:06all, I think that for a long time, I
40:10remember my, my first
40:12job out of college was working at an ad
40:15agency. And I was repeating this as a
40:1723-year-old. I was
40:18trying to tell my boss and my boss's
40:20boss, nobody wants to hear from the
40:23branded Twitter page.
40:25They want to hear from the CEO. They want
40:27to hear from a human being. And that only
40:29proves
40:30more and more true. So I think that
40:32regardless, the human being personal
40:35branding decision is
40:36continuing to accelerate and branded
40:39content just always underperforms.
40:41But also, there's no reason why your
40:44branded content can't repeat all of your
40:47chosen commodity
40:48ideas.
40:49Like if we create, so let's say
40:51TypeShare, our SaaS platform, has an X
40:54account and a
40:55LinkedIn page.
40:57There's absolutely no reason why those
40:59social pages can't just repeat all of my
41:01commodity
41:02ideas.
41:03Now, it wouldn't make sense for TypeShare
41:04to be like, when I was 17 years old, I
41:06was
41:06one
41:06of the highest ranked World of Warcraft
41:08players in North America. Like that
41:09doesn't make sense
41:10because that's a personality data point.
41:12But there's no reason why TypeShare can't
41:14tweet,
41:15here are the five benefits of writing
41:17online. It's a commodity. And it's the
41:18same thing I would
41:19have said. So I actually think that you
41:22get double leverage when you have the
41:25human being,
41:26the founder or an executive team or
41:28whoever, building their own personal
41:31language model that
41:32then the branded content is like, great,
41:33we're just going to ignore the
41:34personality
41:35details and
41:35we're just going to repeat the commodity
41:37ideas, which is what brands do anyways.
41:39You're just choosing which perspectives
41:41do we agree with.
41:43Yeah, and I think you can still, you
41:45know, there's so many brands.
41:47There are a lot of brands like Taco Bell
41:49comes top of mind, which is like, it
41:51feels
41:51like
41:52it's personality driven.
41:54So like, just because it's commoditized,
41:57commoditized content doesn't mean it
41:59can't
42:01be human centric.
42:03I think the future of branded content is
42:08basically not just one human behind it,
42:13but five.
42:15And you can even sign off, oh, it's
42:17Tiffany who did this, or Jeremy who did
42:19this, who wrote this,
42:20and different stories around the brand.
42:23So for example, Taco Bell, let's just use
42:27that.
42:27Instead of just Taco Bell writing
42:30commoditized content around,
42:32or just content around,
42:34oh, hey, we dropped a new crunchy burrito
42:37today,
42:37and here's what it looks like.
42:39It's more around Tiffany, who works at
42:43the Tulsa, Oklahoma store,
42:46has been coming to Taco Bell since she's
42:49been 13.
42:50In fact, her parents met at a Taco Bell.
42:54And then you tell that story.
42:55think about okay so that's a fantastic
42:58example think about how with a little bit
43:01of creativity
43:02if you take or taco bell takes that
43:05approach what what's really happening is
43:08you're you're
43:09opening the door for personality content
43:12and personality content that's that's
43:15what
43:15i was
43:16saying where where lots of like high
43:18personality content often outperforms
43:21more
43:21credible commodity
43:22content. And so I actually think in that
43:26example, Tiffany, the taco czar would
43:29outperform, right?
43:31The other, the other taco company brand
43:34that's just only doing the commodity
43:37version. And this,
43:38this, this is why I think this lens is so
43:40helpful is because it, it forces you to
43:42raise the question
43:43like, what are we actually creating here?
43:46What is the value? And if you're only
43:48creating commodity
43:49content with no personality component,
43:51you have to fundamentally realize it's
43:53like you are creating
43:54the lowest leverage, lower lowest value
43:56form of content in your topic. That's not
43:59to say it has
43:59zero value, but it has the lowest value.
44:02A little, a little creativity means you
44:04can open the second
44:05tranche, which is, well, now we give
44:07ourselves permission to create some
44:09personality content.
44:11And the ultimate goal, the ultimate goal
44:13is to spend enough time crystallizing
44:16your
44:16thinking
44:17through writing on a specific topic where
44:20you ultimately create original content.
44:24And one of the things that I've been
44:26writing about recently is just because
44:29you
44:29wrote it
44:30doesn't mean you own it.
44:32I think that a lot of times people think
44:35because my fingers touched the keys, my
44:38fingers typed
44:39that grouping of words, that means that
44:41grouping of words is mine.
44:43And that's not true.
44:44If you write the sentence, I woke up
44:46today and made coffee, that is not
44:49attributable to you. That is an open
44:51source grouping of words. No one owns
44:53that. And there's this faulty belief that
44:56people have where they think if they're
44:59the ones who wrote it, they're the ones
45:02who did the work. That means they own it
45:04and that means they created something
45:07original.
45:07that is not original thinking you didn't
45:10you didn't think a single new thought you
45:13didn't say
45:14anything that hasn't been said before and
45:17so original thinking is not something
45:19that's like
45:19oh i started writing online and three
45:21days later i had a game-changing insight
45:24original thinking
45:25comes as the result of crystallizing the
45:27way that you see the world in a specific
45:30topic or a small
45:31number of topics for a prolonged period
45:33of time and one of the best examples of
45:35original thinking
45:36Like original does not mean I wrote a
45:39hundred thousand word book on it.
45:41Great example of original thinking was
45:43Pythagoras saying the earth is round, not
45:46flat.
45:47That's six words, like fundamentally six
45:50words that completely changed the course
45:52of humanity that we still associate with
45:55that person.
45:57And you only arrive at those six words
45:59after years and years and years of
46:02crystallizing thinking in a topic.
46:05And so if you kind of loop all the way
46:07back to the beginning here, that's why
46:09there's an irony when people say, I want
46:11to write about everything.
46:14It's like you don't realize that in doing
46:16that, you deprive yourself of ever being
46:18able to go deep enough where you end up
46:20saying something original.
46:22All you're doing is you're just typing
46:24and repeating commodity ideas all day
46:25long, thinking they're yours and thinking
46:27that you have ownership or originality
46:29because you were the one who wrote the
46:31words.
46:31When in reality, those groupings of words
46:33have existed forever.
46:35How do you see writing versus short form
46:38video?
46:40In what sense?
46:42So we talked a lot about content in
46:45general,
46:46on capturing ideas, writing ideas.
46:51And when I hear that, I think of tweets,
46:55I think of blog posts, I think of email
46:57newsletters,
46:57I think of copy on websites.
47:00but on the other hand we have all the
47:03social feeds just hungry for short form
47:07video so my
47:08question to you is how do you think about
47:11writing as it relates to short form video
47:14first of all i
47:15don't one of the most common questions
47:18people ask is which is better which is
47:21better writing or video
47:22which is better podcast or long form
47:25youtube the reality is that the answer is
47:28yes it's all of them
47:29They all work. Me and Dickie, my business
47:33partner, create essentially zero short
47:36form video.
47:39The only short form video we publish is
47:41we have an internal editor and we record
47:43a
47:43long form and he clips out some random
47:46short form and posts them.
47:48Like it is like very low, I'll even say
47:50low value short form. We are not like
47:53intentionally investing in that.
47:55and we run a multi seven figure business.
47:58It's all driven through writing. So like,
48:02yes, there's always a trend or some other
48:05form of content that is working or maybe
48:08working a
48:08little bit better, but that doesn't mean
48:11that the other ones aren't working. You
48:13know, the second
48:14idea is, video is no different than
48:16writing. Everyone uses teleprompters now.
48:19Like when you do
48:21shorts, do you ever script them?
48:24I actually do.
48:27I script them. When I do shorts,
48:30it's like cutouts of
48:33long form content or
48:36I do like animation explainers
48:40that are scripted.
48:42They are scripted? Yeah, those are
48:45scripted.
48:46Either it's cuts or scripted.
48:49Yeah.
48:50So more and more short form video is
48:52scripted and a script is what?
48:55A script is writing and it's the same
48:57fundamental idea.
48:58To make this really tangible, like
49:00walking through the value of the thing
49:03that you're creating.
49:04So pretend I create a short form video
49:07and the first sentence is, here's how to
49:10start a newsletter on Substack.
49:14That is commodity content.
49:16That's not to say that it's not valuable,
49:18but that is fundamentally commodity
49:20content because I am just explaining how
49:22to do something that a lot of other
49:24people
49:25can and probably have explained how to
49:27do.
49:28If we change the first sentence and we
49:30say, I run a $400,000 a year annual
49:33recurring revenue paid newsletter on
49:36Substack, here's how to start a Substack.
49:40Now we've moved up a tranche.
49:42That has a personality detail in it.
49:44It's a little like credibility too, but
49:45that's really a personality detail.
49:47Like I did this, I'm doing this, right?
49:50If we add in the variable, which is sort
49:52of the last variable we haven't talked
49:54about,
49:54which is timely versus timeless,
49:56everything that you create sits on a
49:58spectrum, right?
49:59Timely content has very high short-term
50:02value and very low long-term value.
50:06Timeless content has very low short-term
50:09value and very high long-term value.
50:13And so, for example, I saw you tweet
50:15yesterday or two days ago like an AI
50:17image
50:18meme where you were like,
50:19I'm not waiting for the foldable phone.
50:21I'm waiting for the desktop that's
50:23foldable, right?
50:24So that's a great example of hyper timely
50:27content where you capitalize for 24
50:29hours,
50:29but then the value of that content falls
50:32off a cliff. You can't repost that a
50:34month
50:34from now or three months from now or a
50:37year from now. Right. So this raises a
50:39very interesting question of, all right,
50:42so everything that you create, like how
50:44much value are you actually extracting
50:47and
50:47over what time horizon?
50:49You created a meme, you capitalized for
50:5124 hours, great.
50:53But like, that's all the juice you're
50:55going to get out of it.
50:56You're not going to get any long-term
50:57value out of that.
50:58In contrast, I wrote this essay about AI.
51:02Took me two weeks to write, spent a lot
51:03of time on it.
51:05It got some short-term value, but I'm
51:07going to reference and use that essay for
51:10like the
51:11next 10 years.
51:12And so when you're creating something,
51:14you want to ask, well, where on the
51:16spectrum am
51:16I sitting? And are my expectations
51:19aligned with the thing that I'm creating?
51:22You can't go build
51:23something or write something or create
51:25something that you know is intended to be
51:26very timeless
51:27content and expect it to get the same
51:29short-term distribution as a meme that
51:31capitalizes on an
51:32Apple announcement in 24 hours, right?
51:35There are different expectations. So if
51:37we
51:37tweak the first
51:38sentence of that short form video and we
51:41go, sub stack just launched this new
51:44feature.
51:46I run a $400,000 a year paid newsletter.
51:49Here's how to start a sub stack in 2026.
51:54That word, this just happened or in 2026
51:57fundamentally changes the short, medium
52:01and
52:02long-term value of that content. Cause I
52:05can't repost that in 2030, right? I,
52:07maybe
52:08I could
52:08used pieces of it or I got to rerecord it
52:10or I got to rewrite half of it. And so
52:12when you when
52:13you put all these variables together, you
52:16start to realize like everything you
52:18create in some way
52:19is is adding something to your digital
52:22brain, your knowledge of the world, your
52:25personal language
52:26model. The question is, how much did it
52:29add? How additive was it? If you spend 20
52:32minutes and you're
52:34just repeating the same commodity idea
52:36that you've repeated a hundred times,
52:38maybe you
52:38added like one point of, I said it
52:40slightly differently.
52:41Okay, cool.
52:42We have another like grouping of words,
52:44approved language.
52:45Whereas if you sit down and write a net
52:47new essay about a topic you've never
52:49written about
52:50before and you crystallize a whole point
52:52of view around it, the additive value of
52:54that
52:54to your digital brain is exponentially
52:57higher than the first one.
53:00And so that's my hope with AI is that the
53:02human is doing the writing.
53:03The human's doing the thinking.
53:05The human is doing the choosing of how do
53:07I see the world?
53:08What ideas do I agree with?
53:09Do I disagree with?
53:10What are my personality details, et
53:12cetera?
53:13But the robots are very effective at
53:16repeating and remixing and doing all the
53:20things that
53:21social media managers and even
53:23ghostwriters and VAs have done for a very
53:25long time.
53:26You still got to give the robots
53:29direction around writing in a style that
53:33isn't generic sounding AI.
53:37So now you have touched on the I think
53:39the final piece and it is the hardest one
53:41for people to understand, which is voice.
53:44So what is voice? What is style?
53:48Voice is a bundled term.
53:50when we say someone has a voice,
53:52someone's content has voice,
53:54we think that it's this intangible
53:56quality that can't be replicated.
53:59The reality is voice is really a bundled
54:01term made up of a couple different data
54:03points.
54:04It's like word choice dictates voice,
54:08right?
54:08If I say something is really cool, that
54:11gives you one perspective of who I am.
54:14If I say something is exquisite, that
54:16gives you a different perspective of who
54:18I
54:18am.
54:18And if I say something is fucking dope,
54:20that gives you a different perspective of
54:22who I am.
54:24So individual word choice, like literally
54:26a single word in a sentence can change
54:28your perspective of what the voice of
54:30that
54:30sentence or person is.
54:32Another one is is sentence length and
54:34sentence structure.
54:36if someone writes a paragraph with no
54:39periods and like seven different
54:41sentences
54:42broken up by
54:43semicolons you have a very different
54:46perspective of that person rather than
54:49someone who writes in short terse
54:51sentences with very simple third grade
54:54reading level vocabulary And what
54:57important about this is your takeaway
55:00should be that voice is something that
55:02you
55:03intentionally create
55:05It's not something that just naturally
55:06happens.
55:08And when you become proficient enough at
55:10writing, what's funny, I ran into the
55:12same
55:13thing, ghostwriting for clients.
55:15I will never forget, I had a ghostwriting
55:18client who I sent him a piece and he sent
55:20it back and he was furious and he said,
55:23this doesn't sound like me.
55:25And I looked at it and I went back and
55:27listened to the call and I was like, all
55:28right, he didn't really tell me what it
55:30was.
55:30So I have to guess what he's thinking.
55:31Like, why doesn't this sound like him?
55:33And it occurred to me, oh, I was I was
55:35using the wrong voice.
55:36I made this piece lean too much into
55:39opinion.
55:40And this guy actually spoke referencing
55:42like a lot of studies and a lot of facts.
55:45I didn't rewrite the piece.
55:47All I did was go back in and add in like
55:49two or three sentences that just said,
55:52according
55:52to the New York Times or according to
55:54Harvard Business Review.
55:56I changed 1% of it and I sent it back to
55:59him and he was like elated.
56:01And he was like, oh my gosh, this is so
56:02much better.
56:03It sounds just like me.
56:05And so the irony of writing with AI is
56:07oftentimes AI will give you an output and
56:09people will
56:10look at one single variable.
56:12They'll look at one individual word
56:14choice and go, the whole thing is shit.
56:17And the reality is it's like, no, if you
56:19were proficient, you understood the
56:20qualities that made your voice your
56:22voice,
56:22which most people don't know.
56:24All you would have to do is just go, I
56:26want to trade that one adjective for a
56:28different adjective.
56:29And all of a sudden the sentence sounds
56:31like me.
56:32What do you think of people who don't
56:35like their voice, but who know which
56:38creators' voices they like?
56:41So they might say, you know, I don't
56:43really like my voice, but I really like
56:45how Greg Eisenberg writes his essays on
56:47X.
56:49I truly believe that 99% of the time when
56:51someone says that, they can neither
56:54articulate what makes their voice their
56:57voice, what it is they don't like, and
56:59they can't articulate what makes your
57:02voice your voice.
57:03and usually when people say that it's
57:06more of a mask of they're just afraid to
57:09write period
57:10and so it's a lot easier to point to and
57:13make this like a voice issue the reality
57:15i
57:16have ghost
57:16written for north of 300 different
57:19founders executives investors the vast
57:23majority of the
57:24time i used the exact same voice in the
57:27writing it was authoritative it was
57:30concise it maybe had
57:31like a couple little vocabulary choices
57:34in there specific to that person. But
57:37like
57:3798% of the thing
57:38was just an authoritative thought
57:40leadership style voice. And almost every
57:42one of them said,
57:43this sounds like me. And so if you go, if
57:46you go, I'm not, I'm not joking. And so
57:49if
57:49you go back to
57:50like language is open source, most
57:52groupings of words are commodities. The
57:56reality is that the
57:57voice component in writing is a minority
58:00component. It's a very small piece. It's
58:03not like
58:04this entire piece of writing is an
58:06original voice. Very rarely is that the
58:08case. It happens,
58:09but very rarely. Instead, a piece is 80%
58:12commodity groupings of words with maybe
58:15some subjective
58:17swapping of adjectives and things. And I
58:19think this is one of those, it's like,
58:21you
58:22can't possibly
58:23discover the qualities that you want or
58:26like about your voice without writing?
58:30How are you going to discover, oh, I
58:32actually prefer these types of adjectives
58:34over these
58:34types of adjectives?
58:36You know, I prefer this rhythm in my
58:38writing opposed to this rhythm in my
58:40writing.
58:40And so if you're not willing to do that,
58:43how in the world should you expect AI to
58:47have
58:47any inkling of any of those data points
58:50or choices?
58:50that's the unmade data set yeah i mean
58:53the truth is you know like deep down you
58:56know like it's the
58:57same it's the same concept around like
58:59you know what kind of food you like you
59:01know if you're a
59:01pizza or a pasta person yeah and so like
59:04of course you know the types of words
59:06that
59:07you connect with
59:08how you how how you know your voice it's
59:10just it might be a little deep down you
59:13have to you have
59:14to be a little self-reflective on it, but
59:17it's there. And this is not, to be clear,
59:20this is not
59:21exclusive to the everyday person. I think
59:24oftentimes people think, you know, oh,
59:27I'm
59:27the
59:27everyday person. I haven't achieved this
59:29massive outcome. I bet this is so much
59:31easier for people
59:33who have achieved all these things,
59:34right? They make up all of these
59:36problems.
59:37The reality is
59:38I have I have ghost written for some of
59:40the most like you can accomplish things
59:43any more accomplished than what they
59:46accomplished in the world of business And
59:49even they the core problem was they get
59:52nervous staring at a blank page and they
59:55nervous to put themselves out there
59:59The majority of my value as a ghostwriter
1:00:01was not actually the writing.
1:00:02It was helping the client and giving them
1:00:05permission to put themselves out there.
1:00:09And it was comical to me.
1:00:10ghostwriting for so many people where I
1:00:12would have a client be like, I'm really
1:00:14afraid of putting my
1:00:15perspective out there. And their
1:00:16perspective is like the key to marketing
1:00:18is knowing your customer.
1:00:20I'm like, there's nothing, there's
1:00:21nothing to be afraid of. You are
1:00:23literally
1:00:23repeating a commodity
1:00:24idea that most of the planet agrees with.
1:00:27Like, and so a lot of these are
1:00:29self-inflicted issues.
1:00:32And the, the whole key is like getting
1:00:34out of your own way and hitting publish
1:00:37so
1:00:37that you can
1:00:38discover what these variables are. You
1:00:40know, what makes your personality? What
1:00:43makes your writing
1:00:44style? What makes your voice? The other
1:00:47piece of that is, generally speaking,
1:00:50people are under
1:00:51publishing, like the volume of content
1:00:55that they're putting out is lower than
1:00:58they can actually do.
1:01:00I was having this conversation with our
1:01:03agency, actually.
1:01:06And LCA literally works with the biggest
1:01:10companies on the planet
1:01:13building their new products.
1:01:16So there's, as you can imagine, in the AI
1:01:18world,
1:01:18as you can imagine, that team has just an
1:01:23incredible data set
1:01:25and incredible knowledge base around
1:01:27what's working and what's not.
1:01:29and there's so much content that they
1:01:31could be putting out.
1:01:33And they put out some content here and
1:01:34there,
1:01:35but I think one of the things we were
1:01:37talking about is
1:01:38how do we ship a minimum of five pieces
1:01:41of content a day?
1:01:44And what does that look like?
1:01:45What systems do we need in place?
1:01:47And we could use AI in that process,
1:01:50but what systems do we need in place for
1:01:52that to happen?
1:01:52Because if we're not posting five pieces
1:01:55of content per day,
1:01:56I don't believe we're giving it a fair
1:01:58shot.
1:01:58I would agree. And I would say that it's
1:02:00important for people to remember that
1:02:03that
1:02:03is in many ways, that is the baseline
1:02:06expectation now. And that is just the
1:02:08infrastructure. The question on top of
1:02:11that is, what is the input? If you're
1:02:13posting five times a day, like, where are
1:02:16you pulling the ideas from? Goes back to,
1:02:18well, you need a digital brain, you need
1:02:21a
1:02:21personal language model. Like, where is
1:02:24it
1:02:24all coming from?
1:02:25And the same way, I know you're aware of
1:02:28the trend, you know, you see like OpenAI
1:02:31acquired that podcast, TBPN, and there's
1:02:34a
1:02:34huge move to these like large companies
1:02:37realizing they need a media component.
1:02:39I think that it might be idealistic, but
1:02:43I think that some of the best marketing
1:02:46money that any Fortune 500 company could
1:02:50spend would be plucking really thoughtful
1:02:53journalists and nonfiction writers in
1:02:57relevant or tangentially relevant fields
1:03:00and being like, you have one job.
1:03:03We're going to pay you a fat salary and
1:03:05your only job is to think deeply about
1:03:08our
1:03:08industry and write long form content
1:03:10about
1:03:11it.
1:03:12That's it.
1:03:12Because what's really happening is it's
1:03:15not that like that individual piece of
1:03:17writing is the silver bullet and that's
1:03:19going to drive all the sales and they
1:03:21make
1:03:21it all super promotional.
1:03:23What's really happening is you have a
1:03:25very smart person who's paid to think
1:03:27deeply about a topic, hopefully over time
1:03:30creating net new original thinking and
1:03:32ideas that get associated with the
1:03:34company.
1:03:36And that library of content feeds the
1:03:38infrastructure.
1:03:40It feeds your whole supply chain.
1:03:41It's like just to make it visceral, Greg,
1:03:43it's like imagine you were like, cool,
1:03:45I'll pay you a million bucks a year.
1:03:46I want you to do nothing but think deeply
1:03:48about the future of AI and technology and
1:03:50the implications it's going to have on
1:03:51product building.
1:03:52And I just want you to build like a world
1:03:55class long form content library that all
1:03:57of our different departments can pull
1:03:59from.
1:04:00The robots do the repeating.
1:04:01We can pluck short form.
1:04:02We could turn it into automated videos
1:04:04like all the infrastructure.
1:04:06I just want you to create the highest
1:04:07quality input.
1:04:09Now, imagine if Salesforce did that with
1:04:12like a Pulitzer Prize winning journalist
1:04:15or Slack did that.
1:04:16I think that the leverage you get out of
1:04:19that is so it's so hard for people to
1:04:21wrap
1:04:21their heads around.
1:04:23And they're so focused on the minutia of
1:04:25like we got I post five times a day and
1:04:26what time of day and how do we go viral
1:04:28and can we pay for retweets and like all
1:04:30none of it matters.
1:04:32None of it matters.
1:04:33What matters is you have the
1:04:35infrastructure and you have a very
1:04:37valuable content library that you pull
1:04:39from and leverage for a long period of
1:04:41time Yeah I mean that why people are
1:04:43starting to say the company brain is the
1:04:45moat
1:04:46Yes.
1:04:47Right?
1:04:48Because if I hired you, Cole,
1:04:53and you're adding to this company brain
1:04:54and it's getting more and more valuable
1:04:56and I can use robots to go and slice and
1:04:58dice it a million ways,
1:05:00I know it sounds crazy to pay someone a
1:05:02million dollars,
1:05:04But it's not that crazy if the value from
1:05:07that million dollars is thousands of
1:05:10pieces of content.
1:05:12It's improving workflows.
1:05:13It's helping with the product.
1:05:15It's pretty insane, right?
1:05:17And that's why you're starting to see
1:05:19actually companies spend millions of
1:05:21dollars, tens of millions of dollars.
1:05:23And I think it will be hundreds of
1:05:25millions of dollars on data licensing or
1:05:28buying data.
1:05:29I saw Micro One buying data, Spirit
1:05:31Airlines data, for I think millions of
1:05:34dollars.
1:05:36And why? Because they're going to use
1:05:38that data.
1:05:39They're going to mine that data.
1:05:40And I think that's where everything is
1:05:44going.
1:05:45That's exactly it.
1:05:47It's just people understand this idea
1:05:49more in technology.
1:05:51They go, oh, we acquired a company that
1:05:53just filed for bankruptcy and we bought
1:05:56all their data.
1:05:57and everyone's like, yep, I instantly
1:05:58understand the value.
1:06:00The reality is language and writing
1:06:02and libraries of content are data.
1:06:06And you are buying groupings of words,
1:06:09you are buying ideas, you are buying
1:06:10crystallized thinking
1:06:12that you can leverage the same way you
1:06:13leverage data.
1:06:15Yeah, it's funny because I bet the Spirit
1:06:18Airlines people
1:06:19are probably like, ha ha ha ha,
1:06:21I can't believe they bought our data for
1:06:22millions of dollars.
1:06:24Idiots.
1:06:25and the Micro One people are like,
1:06:28ha ha ha ha, I can't believe they sold us
1:06:30the data for so cheap.
1:06:32Yeah, well, it's always a question of,
1:06:34it's actually not really the value of
1:06:37those assets in and of themselves.
1:06:40The question is, how do you leverage it?
1:06:44If you're a company and you buy some
1:06:46writer's entire catalog
1:06:49and library of content,
1:06:50and then you don't have the operational
1:06:52infrastructure
1:06:52to scale it and squeeze all the juice out
1:06:54of it,
1:06:55then that was a horrible decision
1:06:56because you're not able to do what you
1:06:58should do with it.
1:06:59But if you have a really smart person as
1:07:02your input
1:07:02for your entire content marketing
1:07:05machine,
1:07:06like literally one person doing that
1:07:09for Slack or Salesforce or Airbnb or
1:07:12whoever,
1:07:14one smart person could build
1:07:16the entire digital brain library of
1:07:19content
1:07:20that their entire marketing machine uses
1:07:23and reswizzles and remixes and posts and
1:07:25chops and
1:07:26cuts and paste. To me, that's the future.
1:07:28I mean, we're out of time, but that's
1:07:30like
1:07:31a whole other
1:07:32startup idea that someone needs to do,
1:07:35which is just do company brains as a
1:07:37service. Come on.
1:07:41It's a home run. If I was excited at all
1:07:43about B2B, I would go do it, but I'm not.
1:07:46Cole, you've been so generous with your
1:07:49ideas, the whole playbook. You open
1:07:51sourced it. I really
1:07:53appreciate it. Where could people get to
1:07:55know you better, find you, and get more
1:07:57of
1:07:58your ideas?
1:07:59Always, always pumped to chat, my man.
1:08:01You can find me on X. I still want to
1:08:03call
1:08:03it Twitter.
1:08:05You know, I'm an old head. You can find
1:08:07me on X. And if anyone wants to build
1:08:10their digital brain
1:08:11and library of content, you should come
1:08:13join us for the next Ship 30 for 30
1:08:14cohort. We're going to
1:08:15do that in October. Should be a lot of
1:08:17fun.
1:08:17Cool. Yeah, I'll include links
1:08:19for all that in the show notes,
1:08:21in the description. Cole, you've
1:08:23been on the pod probably, what,
1:08:25three, four times? Is it three?
1:08:27Three or four? Three? Yeah.
1:08:29I love having you on. I hope this episode
1:08:31is well-loved.
1:08:34It's one of those things where
1:08:35this is like, you're talking about
1:08:37timeliness, timely versus timeless,
1:08:39and this is a timeless episode.
1:08:41People who
1:08:43ingest these ideas
1:08:452026 now like 2029 2032 like these are
1:08:48these are ideas that are always going to
1:08:52to be so so valuable and it's it's sort
1:08:55of in some ways like an eat your
1:08:58vegetables episode
1:09:00yeah you know versus like drinking a
1:09:03smoothie there's just so much nutrients
1:09:06in
1:09:06this and i do
1:09:07believe that people who actually you know
1:09:09if you've made it here the people
1:09:11actually
1:09:12who who've made
1:09:12here, they are going to understand how to
1:09:15have strong points of view, how to use AI
1:09:18to
1:09:18create some of this content, how to think
1:09:20about how it all plays together.
1:09:22And they're lucky.
1:09:25Well, thanks, man.
1:09:26Yeah, that's my goal.
1:09:27As timeless as possible.
1:09:30I'll catch you later, Cole.
1:09:31Thanks, everyone.

About this transcript

This page contains the full transcript of $30M Writer: Never write AI slop again by Greg Isenberg, generated from the public captions YouTube serves with the video. The transcript has 11,398 words across 2,055 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.