How to Make AI Sound EXACTLY Like you — Transcript
Full transcript
- 0:00Everyone loves using AI to write.
- 0:02Emails, messages, video scripts, but we
- 0:05all hate getting AI slop. [music] And we
- 0:07all know what it sounds like. Where it's
- 0:09like delve and quietly in that thing
- 0:11where every sentence goes not this, not
- 0:13this, but that. I promise you, we can
- 0:16all tell. I've written a best-selling
- 0:18book and I'm actually writing my next
- 0:19one now. And as much as I love using AI,
- 0:22it can't replace a human where it really
- 0:24matters. But for everything else, I
- 0:27found a way to make AI think like me,
- 0:29write like me, sound like me. And today
- 0:31I'm going to show you exactly how to do
- 0:32the same thing. And be sure to stick
- 0:34around because at the end I'll show you
- 0:35how to make AI sound like you in
- 0:38everything you do, not just writing. And
- 0:40it all starts with building the thing
- 0:42that's going to hold it all together.
- 0:43Before AI can sound like you, it needs
- 0:46something to work with, a container to
- 0:48live in. We have to build the brain.
- 0:53I started creating content 15 years ago.
- 0:56And when I started, I had no idea what I
- 0:58was actually building. See, [music] 15
- 1:01years of everything I've ever learned,
- 1:02my knowledge, my ideas, my stories, my
- 1:05preferences, everything was right there,
- 1:08ready to be mined. If I had to build
- 1:10that from scratch today, it would take
- 1:12me years of talking or documenting to
- 1:15pull that information out. And that's
- 1:17the part that nobody tells you. The
- 1:19value isn't in one thing you record
- 1:21today, it's in the 10 years of stacked
- 1:24up, documented, cataloged information.
- 1:27And I get it, if you haven't documented
- 1:29anything so far and you're just getting
- 1:31started, this is where we start. [music]
- 1:33We need to record.
- 1:35Okay, we need to save information. We
- 1:39need to find all the documents that
- 1:41we've created in the past. Think about
- 1:43like all the playbooks, everything
- 1:45you've ever written. It might be in
- 1:46other emails from the past on old hard
- 1:48drives. The more original content that
- 1:51you created, that you can point to, the
- 1:54better the AI is going to be. It's the
- 1:56system that this whole thing is going to
- 1:58run on. And the more we give it of
- 1:59things that you've actually captured
- 2:01that are you, that relevant context will
- 2:05build the most powerful brain you can
- 2:06create. And look, I did a whole video
- 2:09about how to structure your AI brain,
- 2:10and I'll link it at the end, but I just
- 2:12want you to know, if you have not
- 2:14started this process, we have to build
- 2:16the brain. Here's how you can do it, and
- 2:18it's super simple. First off, you have
- 2:20to record everything. I'm recording
- 2:22every meeting, I'm recording all my
- 2:24calls, I'm recording all my decisions.
- 2:26I'm even recording every AI chat that I
- 2:28have in all my different AI systems. I
- 2:31pull all of that together. That becomes
- 2:33the context for how I think, how I make
- 2:36decisions, my frameworks, my current
- 2:38status in my life. And look, there's a
- 2:39couple awesome tools. I use Granola for
- 2:42a lot of my meetings. A lot of people on
- 2:43my team have Pocket, which is a physical
- 2:45device that they can just record from.
- 2:46I've others that use like WhisperFlow,
- 2:48and they just record voice notes and do
- 2:50dumps at the end of the day with like
- 2:51where they're at and what's going on.
- 2:53See, the tool doesn't matter, it's the
- 2:55capturing that does. The other day I was
- 2:58doing a retreat with my wife, and we
- 2:59were driving, I said, "Pull out your
- 3:01voice memo app on your iPhone, and just
- 3:03record us talking." And we did a full
- 3:05dump. We asked ourselves, "In 5 years,
- 3:07what will be true? Let's talk about it.
- 3:09What do we want to accomplish?" We
- 3:10captured [music] everything over a
- 3:1227-minute conversation while we're
- 3:13driving, and then we grabbed that and
- 3:16gave it to the brain. And it parsed
- 3:17through the different voices and said,
- 3:19"This is Renee's task list. This is
- 3:20Dan's task list." And then we started
- 3:22executing. You just got to make sure
- 3:24that as soon as you have that thought or
- 3:26you made that decision, it gets out of
- 3:28your head and it gets documented. Your
- 3:30brain is a horrible place to try to
- 3:32remember things. Get it out of your head
- 3:34or you get in trouble. And then the key
- 3:36is, you have to be sure that you give it
- 3:38one home base. I like to use Obsidian,
- 3:41you can use Notion, you can use a Google
- 3:43Drive. It doesn't matter actually what
- 3:45you use, you just have one place where
- 3:47everything [music] gets dropped. Because
- 3:49that's what we'll be referencing as we
- 3:51build up this humanization system for
- 3:54your AI. And look, I'll give you my
- 3:56whole prompt to ingest all the
- 3:58information I discuss into the brain in
- 4:00the description below. You can just copy
- 4:01and paste it and it'll answer questions
- 4:03and then it'll get you started. [music]
- 4:05If you do this for 30 days, you'll have
- 4:07more raw material than most people build
- 4:09in a year because [music] you're
- 4:10captured. This is cool. You have
- 4:13everything captured, raw material,
- 4:15perfect.
- 4:16But having a bunch of transcripts kind
- 4:18of useless unless we teach the brain how
- 4:21to think. So now we got to make sure we
- 4:23make it think like you.
- 4:27This is actually my favorite part about
- 4:29AI. 2 years ago, I remember my team
- 4:32exploded and I just kept getting asked
- 4:34the same questions from different
- 4:36people. Things I've answered, things
- 4:38I've written books about, things I had
- 4:39videos about, but I realized the context
- 4:42and the timing were not matched. Reading
- 4:44a whole book to get one answer, that's
- 4:45complicated. Not only that, there's
- 4:47context that might be missing that the
- 4:48person has around the question that
- 4:50would change my answer. It was crazy cuz
- 4:53I was like
- 4:54could I build an AI system that could
- 4:57answer those questions from those people
- 5:00as good as me?
- 5:02Better than me? Turns out you can. So I
- 5:05designed this new system called Dan AI
- 5:08and that one tool allowed me to buy back
- 5:11literally tens of hours a week because I
- 5:15had so much documented already. See, the
- 5:18key if we don't do it like this is that
- 5:20your brain will just hallucinate. It
- 5:21will fill in the gaps. It won't actually
- 5:23answer the question as you would, but
- 5:25the more information we have in the
- 5:26brain and the way we teach it to think
- 5:28like us, the better the responses are
- 5:29going to be. So look, now that we've
- 5:31built this foundation to capture some of
- 5:34the information, you're now recording
- 5:36things, we have to give it more
- 5:37information so that it can learn to
- 5:39think like you. I'm talking email. I'm
- 5:41talking Slack. I'm talking every
- 5:43document you've ever created because
- 5:45within those documents are embedded into
- 5:48it your beliefs. Essentially, what I
- 5:51call your unique points of view, the
- 5:52stories that drove those beliefs. If I'm
- 5:55just giving you an answer, that's one
- 5:57thing. If I give you the answer with the
- 5:59story and the principle and a visual
- 6:01model, now it's showing people how you
- 6:03think. So, how do we teach the AI the
- 6:05way you think? What we need to do is we
- 6:07have to plug our AI into all the current
- 6:10tools. These are called connectors.
- 6:12Basically, give AI access to your Gmail,
- 6:15to your Google Drive, to your Notion, to
- 6:17your Slack, to your granola.
- 6:19Essentially, as many different
- 6:21connectors as you can, it'll teach it
- 6:23how to make the connections for how you
- 6:25think. And the truth is, each one takes
- 6:27about 2 minutes. And look, you might
- 6:29feel super behind, but you're not. The
- 6:32other day I had 40 top-level CEOs, okay?
- 6:34These are people making eight figures a
- 6:36year, and I asked them to analyze their
- 6:38inbox so that we could build a tone
- 6:40document, something I'm going to show
- 6:41you in a second. And a third of the room
- 6:43had never connected their email to their
- 6:46AI chat. There's so much power in that.
- 6:49Do you know that I hardly ever search my
- 6:51inbox through the inbox tool? I always
- 6:54use the AI. It finds things faster,
- 6:56better, and more accurately than the
- 6:58existing search features in the tools
- 7:00themselves. And pro tip, if you create
- 7:02content just like me on Instagram,
- 7:04TikTok, YouTube, etc., point to those
- 7:06sources as well because it can use that
- 7:08information to teach itself how to
- 7:10think. Now, that'll get you a lot of
- 7:12information, but there still might be
- 7:13things that it feels like there's gaps
- 7:15around its thinking or its knowledge.
- 7:17So, you can ask it to interview to fill
- 7:19in those gaps so that it can save it in
- 7:21the brain so that it uses that when it's
- 7:23trying to think like you. So, set 20
- 7:25minutes aside and answer in your voice
- 7:27as it needs to fill in those gaps.
- 7:29Here's the prompt you run. I want you to
- 7:31have all the context you need to think
- 7:33like me. Use all the tools I've
- 7:35connected together as much as you can
- 7:37about how I think, what I believe, and
- 7:39what I've done. Then, interview me to
- 7:41fill the gaps, my points of views, my
- 7:43contrarian takes, the stories behind my
- 7:45decisions. Ask at least four rounds of
- 7:47questions, and when you're done, output
- 7:49a document called your name brain system
- 7:52that explains all this simply. Save that
- 7:54doc. Every AI system you use going
- 7:56forward will start with it so that it
- 7:58can think just like you. Now it thinks
- 8:00like you, but the problem is it still
- 8:02sounds like AI slop. So how do we fix
- 8:05that? Did you know a marketing survey
- 8:07consumers two years in a row in 2025 20%
- 8:11said heavy AI use made them trust a
- 8:13brand [music] less. 2026, it's up to
- 8:1640%. It's doubled in a single year. And
- 8:19guess what? Gen Z, the ones that grew up
- 8:21with AI, punish it the hardest. That's
- 8:2354%. They hate it. People buy
- 8:27from people they know, like, and trust.
- 8:30And the problem is when the people go
- 8:31away, people don't know how to buy.
- 8:34Kylie Jenner's brand has like a thousand
- 8:37time more followers than her cosmetic
- 8:39company. Nobody's following a business,
- 8:41especially if they think that AI created
- 8:43all the marketing content. Nobody's mad
- 8:45at you for using AI. They're just mad
- 8:47because you made it sound like AI. And
- 8:49they just want to know that a real
- 8:51person was involved in the creation of
- 8:53the output. We have to make it sound
- 8:58like you.
- 9:00This is probably my favorite party
- 9:02trick. Okay, I'm going to show you how
- 9:04to do it, and once you see it, you're
- 9:05like, "Oh my gosh, I've been doing this
- 9:06hard." Here's how we train the AI to
- 9:08sound just like you. First off, you have
- 9:11to feed it your real writing. You
- 9:13probably email completely different than
- 9:15you post in Slack, and you probably text
- 9:17message with your friends, or even how
- 9:19you talk on a camera. So we need to set
- 9:21it up and give it the data so that it
- 9:23can actually learn how you talk in those
- 9:24different containers. Then what I did is
- 9:26I took those buckets, those individual
- 9:28outputs, and I asked the AI to analyze
- 9:31it to find the patterns. Look at the
- 9:33sentence length. Look at the
- 9:34punctuation. Look at my favorite words.
- 9:36Look at my capitalization. Look at how I
- 9:37open. Look at how I close. Look at how I
- 9:39sign off. If I ask you read out loud
- 9:42your last 20 emails, you could probably
- 9:44see a pattern about how you email that's
- 9:46different than somebody else. The AI
- 9:48knows how to look for that. And if you
- 9:50haven't captured this, make a list of
- 9:51words that you would never use or
- 9:52punctuation that we you never use and
- 9:54just like give all of that to the AI.
- 9:56Things like delve and leverage and
- 9:58robust and quietly. These are things
- 10:00that humanity has said a lot and that's
- 10:03why it shows up in the models. But you
- 10:05may have never used the word quietly in
- 10:07any email you've ever written. Here's
- 10:09the prompt I want you to run. Super
- 10:11simple but clear. I want you to write in
- 10:13my voice so well that nobody can tell AI
- 10:16was involved. Pull my 20 most recent
- 10:18sent emails, my 20 most recent Slack
- 10:20messages, and my 20 most recent
- 10:22documents. Break down each one
- 10:24separately. The length of my sentences,
- 10:26the words I use, the words I never use,
- 10:29how I open, how I close, how I sign off.
- 10:32Keep each medium separate. So my email
- 10:34voice is different than my Slack voice
- 10:36and don't blend them together. If
- 10:37something isn't in the samples, ask me
- 10:39instead of guessing. Then output a
- 10:41document called voice style guide that I
- 10:44can paste into any future chat. And if
- 10:46you have other things, like I went
- 10:47through and screenshotted all my text
- 10:49messages and gave it that and said,
- 10:50"Hey, create a section for my text
- 10:52messages." I went and analyzed a bunch
- 10:54of outlines of videos and transcripts of
- 10:55videos and gave it that. So that if I'm
- 10:57writing scripts, I want the bulleted
- 10:58done in the way that I talk. Turns out
- 11:01how I talk on camera is completely
- 11:02different than how I text to my friends.
- 11:04Even the punctuation's different. Like
- 11:06when I text, I don't punctuate. When I
- 11:08write emails, there's a little bit more
- 11:10formal. When I'm replying to team
- 11:11members on Slack, I know that I've got
- 11:13to be super crystal clear because
- 11:14everybody's going to take the time to
- 11:16read that post and I need to make sure
- 11:17it's clear. I have never used an emoji
- 11:19in a script. You probably have never
- 11:22either, but you would in a text message
- 11:23reply. So now you've got two documents,
- 11:26right? You've got your brain system and
- 11:27you've got your style guide. Both go
- 11:29into every chat going forward. It will
- 11:31sound like you, but I need to encourage
- 11:34you, you still need to review it. I use
- 11:37it to generate ideas, to help me draft,
- 11:40to come up with concepts, but I'm still
- 11:42going to review it to make sure it's
- 11:44great. I use it for research. I ask it
- 11:46to give me options of how I would say
- 11:48things three different ways based on
- 11:49other things that I've written about.
- 11:50I'm looking for ways to get leverage,
- 11:52but it's still me reviewing and editing.
- 11:55I would never let something go out that
- 11:57was written by AI that makes no sense to
- 11:59me. But will I get to a draft faster?
- 12:01100%. Remember I said earlier there's a
- 12:03bunch of like banned words in my list of
- 12:06things? It's because I started noticing
- 12:08that all AIs kind of follow the same
- 12:10pattern, and if you could just tell it
- 12:12to never use these terms or this pattern
- 12:14or this structure, then it'd be
- 12:16impossible for anybody to think that it
- 12:18was AI slop. So, if you want the full
- 12:20list of banned terms and words and
- 12:22structure, just go find me on Instagram,
- 12:24follow me, and message me the word
- 12:25YouTube human, and I'll send it over.
- 12:27So, now we got our brain. It knows how
- 12:29to think like you. It knows how to sound
- 12:31like you. But this is where things get a
- 12:33little bit dangerous. Here's the thing.
- 12:36The better it gets, the more dangerous
- 12:38it is for you to make a huge mistake. I
- 12:41feel this every day. People know me as
- 12:44the AI guy. So, when I'm in my DMs and
- 12:47I'm replying real thumbs, people just
- 12:49assume it's a bot. Whenever I
- 12:52post a video, you look at the comments,
- 12:54AI, AI, AI. What? I can't know how to
- 12:58wake surf and mountain bike and run
- 13:01companies and be that like why do you
- 13:03guys think it's all made up? But I did
- 13:05it to myself because I teach people how
- 13:08to use AI to get more leverage to buy
- 13:10back their time. But the most human
- 13:11thing I can do is be the person that
- 13:14talks to the people, to show up here,
- 13:16real camera, real real flesh, real skin,
- 13:19real real markers, real pieces of paper
- 13:22and pen, because I realize this is the
- 13:25connection. This is the best use of my
- 13:27time. This is the moment of leverage. AI
- 13:30is just a tool that makes me faster. It
- 13:32allows me to shoot videos every week. I
- 13:34look at all the places where there's
- 13:36bottlenecks, but a human involved, me
- 13:38involved, makes it human. Think about a
- 13:41great song. It's not the song or the
- 13:43beats, it's the it's the vibe of it.
- 13:46When Taylor Swift writes a banger, she
- 13:48doesn't even always know how it's going
- 13:50to work, but once she hears it, she
- 13:51knows it. That is taste. Even somebody
- 13:54else singing Taylor Swift's song on the
- 13:56same stage wouldn't have the same
- 13:58impact. There's something about her,
- 14:00that spark that everybody loves. People
- 14:02love her humanity behind it. The
- 14:05contradiction, the chaos, the the lived
- 14:07experience, the story, the narrative.
- 14:10That is not AI, that is human. If she
- 14:13put out an AI song, I guarantee the next
- 14:16day album sales would fall. Nobody would
- 14:18trust her anymore. It takes a whole life
- 14:21to build a reputation. It takes one
- 14:22moment to lose it all. Now you have
- 14:24something that that sounds exactly like
- 14:26you, even thinks just like you. The
- 14:28mistake would be for you to stop
- 14:30thinking, for you to stop involving
- 14:32yourself in creation. MIT hooked up 54
- 14:35students writing essays and watched
- 14:36their brains. The AI group had the
- 14:38weakest brain activity in any group.
- 14:40It's like they deferred their thinking
- 14:42to this device. It's why most people
- 14:44can't remember phone numbers anymore.
- 14:46Even in navigation, using Google Maps
- 14:48makes you not use your brain as much, so
- 14:50you don't remember things. In that
- 14:52study, 83% couldn't quote their own
- 14:55essay minutes after writing it. Think
- 14:57about that.
- 14:58They wrote it, and then they couldn't
- 15:00tell you what was in it because they
- 15:02actually never thought it. They just
- 15:04watched it appear. There's a difference.
- 15:06What you want to do is be at the
- 15:08inception of every idea. And then we use
- 15:11AI to get it to the world as fast as
- 15:13possible. At the top of everything, we
- 15:16want to make sure you are there. You
- 15:19need to make sure that it's working. You
- 15:21need to make sure that it's thinking.
- 15:22You need to make sure that the brain is
- 15:24built in a way that represents you.
- 15:27Boom, at the top.
- 15:29Here's a crazy question. Would you watch
- 15:31F1 if you knew that every person in
- 15:33there was a computer or an AI? No. The
- 15:36human needs to be part of the narrative.
- 15:38The human part of the process is what
- 15:40you love. It's why you're involved. You
- 15:42don't just want a bunch of AIs talking
- 15:44to each other and it's like, "Who's even
- 15:45running this company anymore?" So,
- 15:48my number one rule
- 15:50is that we have to use the AI to pull
- 15:54from you, not push onto you.
- 15:58Okay? Pull. Pull the ideas from you. You
- 16:00are what makes you unique. You are the
- 16:02thing. Don't let it tell you how you
- 16:04should be or how it should act. Make it
- 16:06interview you. Instead of you answering
- 16:08it or having it tell you what you should
- 16:10do in your business, you tell it. Here's
- 16:12what I'm thinking.
- 16:13Help me align with this. If you need to
- 16:15write a newsletter, tell it, "Here's
- 16:17what's on my mind lately. Here are the
- 16:19tools that I'm working with. Go analyze
- 16:20everything that I've been working on and
- 16:22give me some ideas for it. But don't let
- 16:24it tell you what you should publish. You
- 16:26got to look at it and decide, "That's
- 16:27something I can back." AI is just a tool
- 16:30and you have to treat it like one. Just
- 16:32like you don't ask a hammer how to build
- 16:34a house. You decide what kind of house
- 16:36you want to build. The hammer is just a
- 16:37tool. So, for example, here's how I
- 16:39prompt it. I need to write a newsletter
- 16:41about X. That is what I come to, the X.
- 16:44I want you to use the tools I've
- 16:46connected to gather the most context
- 16:47around this topic. Then interview me to
- 16:50get my thoughts on the matter. Use my
- 16:52brain system and my voice style guide to
- 16:54make it sound exactly like me. You had
- 16:56the recipe books, now you need to feed
- 16:58it the ingredients.
- 16:59So, now you have the playbook. How to
- 17:01humanize AI and allow you to get more
- 17:03done. But the truth is, everybody's got
- 17:06the same tools. They got access to the
- 17:07same models, the same prompts, the same
- 17:10videos, the same speed of execution.
- 17:13What's left and who wins is actually
- 17:15what you think. It's a decision to
- 17:18invest in you, to invest in your
- 17:19reading, to invest in your skills
- 17:21development, to invest in your habits,
- 17:22to continue to grow that muscle of
- 17:24learning and teaching and applying. If
- 17:27you do that, you will develop this part
- 17:29of your brain. If you just defer to AI,
- 17:31you will wake up and wonder how did
- 17:33everybody run circles around you. So, I
- 17:36want you to make a commitment below in
- 17:37the comments and let me know out of
- 17:39everything I shared with you, what's one
- 17:40thing you're going to keep as you, the
- 17:43human part you, that you would never
- 17:44have somebody else do. Maybe it's going
- 17:46to networking events. Maybe it's doing
- 17:48sales calls. I don't know. You tell me.
- 17:50I'm just curious. What part do you love
- 17:52to do that you're going to keep as you
- 17:54and we use AI to automate and do all the
- 17:55rest so you don't have to do it. And
- 17:57remember, if you want my full list of
- 17:58words and phrases and things that it
- 18:00should never use, punctuations, just go
- 18:02find me on Instagram and DM me the word
- 18:04YouTube human after you follow me and
- 18:06I'll send it over to you. And if you
- 18:08want to learn how to build and structure
- 18:09your own AI brain, click here and I'll
- 18:11see you on the other side.
About this transcript
This page contains the full transcript of How to Make AI Sound EXACTLY Like you by Dan Martell, generated from the public captions YouTube serves with the video. The transcript has 3,796 words across 520 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.