YT YouTube2Text

YouTube2Text → Transcripts → How to Use Claude To Grow From 0 to 100K Followers (Full Course)

How to Use Claude To Grow From 0 to 100K Followers (Full Course) — Transcript

by Kallaway · 11,863 words · 1,647 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00In this video, I'm going to explain how
0:01to use Claude to grow faster on any
0:04social media platform. And this will
0:05cover the entire content workflow from
0:07strategy and topics to formats, hooks,
0:10scripts, [music] edits, and systems. If
0:12you wanted to learn how to use AI to
0:13make better content faster that isn't
0:16slop, this is going to be a master class
0:17for you. By the way, if you don't know
0:19me, my name is Callaway. I have millions
0:20of followers. I've done billions of
0:22views. And content is all I do all day
0:24long. All right. Now, here's the
0:25workflow we're going to be running
0:26through in this video. I call it the
0:28content growth system. First, we start
0:30with strategy and positioning. There are
0:32five cloud skills that I think anyone
0:33making content should run in succession.
0:35So, we're going to walk through that
0:36whole thing. And then from there, we're
0:37going to get into the actual workflow to
0:40make a video itself. You go from topic
0:42to research to format to hook, script,
0:44record, edit, that whole flow. We'll
0:46talk about the AI placement within that
0:48flow and every skill you want to run.
0:50And then lastly, after you go through
0:51that workflow is the learning loop to
0:53actually take the winners, learn from
0:55them, and then reinforce that back in.
0:57We call that the content coach. So,
0:58we'll go through that at the very end.
1:00Now, to run this whole flow, all we need
1:01are two things. We need the Claude
1:03desktop app, and we need the
1:04Sandcastles.ai MCP, the social media
1:07plug-in installed for Claude. I'm not
1:09going to go through the full
1:10installation process for how to set this
1:11up. I've got videos linked below. It
1:13takes like 3 minutes. It's very easy to
1:14do, but very simply, the reason we need
1:16this is because Sandcastles is the data
1:19layer for social media. It allows us to
1:21deep analyze any video from our channel,
1:23our competitors, map the niche, get the
1:25transcripts, and that really is the
1:26engine that allows us to unlock these
1:29hidden insights for how we should twist
1:31the strategy. MCP is just a fancy word
1:33for plugin. It's the bridge that allows
1:35you to get the social media data into
1:37Claude. If you don't have the social
1:39media data in Claude, that's where you
1:41get AI slop. Claude is just guessing. It
1:43has no context on you. It doesn't know
1:45what videos are working. Doesn't know
1:46how the competitors are posting. And so
1:48it just generates absolute trash. But
1:50once you have the social media data
1:52piped in from your channel and your
1:53competitors and it is deep analyzed to
1:56give to the clawed brain, that's where
1:57the magic happens. This combination of
1:59Claude and Sandcastles, the data loop is
2:01really why I've been able to grow so
2:03much faster than most people and why the
2:05business owners that I coach directly
2:07are crushing in their niche. They're all
2:09running this learning loop. So at this
2:10point in the video, I'm going to assume
2:11you have the Claw desktop app and you've
2:13installed the Sandcastles.ai MCP plugin.
2:16By the way, for everything we do, we're
2:17going to run with Fable 5. That is the
2:19most expensive but the best model in
2:20Claude. You can probably run with Opus 5
2:23as well. I just want the best insights
2:24we can get. Now, let's start with the
2:25content strategy and positioning piece.
2:27Like I mentioned, there are five
2:29critical Claude skills I recommend every
2:31single person run. I will link them
2:33below. You can run them in succession.
2:34You've got the niche interviewer, you've
2:36got the unique positioning builder,
2:38you've got the bullseye builder, you
2:40have the topic scanner, and the format
2:42finder. Now, for this part of the video,
2:43I'm going to pretend like I'm a plastic
2:45surgeon. I'm not going to be the social
2:46media growth guy. I'm not going to be
2:47the tech guy. I'm a plastic surgeon in
2:49LA that's trying to use content to drive
2:52leads and clients for my practice. Call
2:54me Dr. Callaway. And you can tell I'm
2:56obviously not a real plastic surgeon. I
2:58would have fixed this nose a long time
2:59ago. Now, this process is what I
3:01actually run on both sides of my content
3:03world. But I want to prove that anybody,
3:05even if you don't know what you're doing
3:07with content, can run this workflow and
3:09get insane insights. So that's what
3:11we're going to do, Dr. Callaway, for
3:12this. Now, to start this process off,
3:14all you have to do is go into Claude, go
3:15to new chat, make sure Fable 5 is
3:17selected, you paste in this prompt, and
3:19I'll explain this in a second, and you
3:21drag in these six skills. So, I'll have
3:24these linked. You guys will be able to
3:26do this. We'll drag them in. So, we have
3:28those six skills in here. So, we have
3:29the growth orchestrator, kind of like
3:30the growth system, the master skill, and
3:32then we have the first five that talk
3:34about the strategy and positioning like
3:35I mentioned. Now, the prompt very simply
3:37is I want to run the content growth
3:38system. I am a insert your avatar here.
3:41So, I'm a plastic surgeon in LA, Dr.
3:43Callaway, that wants to use content to
3:45drive leads and clients for my practice.
3:47You have this. I would recommend running
3:48this in co-work. You've got Fable 5
3:50selected, and we press go. So, now that
3:52is kicking off, and I'm going to step
3:54you through each piece of this flow. All
3:56right, first up is the niche interviewer
3:58skill. This is very critical for
3:59everyone to start with. Honestly, if you
4:01make content, if you're a marketer,
4:02creator, business owner, founder, and
4:03you've never been interviewed by Claude
4:06to kind of contextualize your
4:07positioning for content, I recommend
4:09doing this. the skills obviously linked
4:10below for free. This will be five to six
4:12questions that'll interview you running
4:14through a framework that I know works
4:16for positioning and you can kind of
4:17answer these questions, help set up the
4:19framework for this whole system. So,
4:20what this will do when you're done with
4:22this super quick 5-minute interview,
4:23it'll give you the answer to who is your
4:25ideal avatar, what is your target
4:27authority statement, what are the pain
4:29points of that avatar, what's their
4:30dream outcome, and ultimately what is
4:32your potential competitive
4:34differentiation? You giving those
4:35answers will help this engine a lot,
4:37right? So, I'm going to go through what
4:39that looks like for me, Dr. Callaway, as
4:41the plastic surgeon. So, just at a high
4:42level, this is kind of what I wanted to
4:44do. We serve upscale clients, typically
4:47women between 45 and 55. These are
4:49people who have spent money on facial
4:52filler or Botox or other smaller
4:55procedures, never done a full plastic
4:56surgery run. They're very concerned
4:58about the procedure not working super
5:00well. They obviously want it to look
5:01great, and they're concerned about their
5:03aging. They have kind of upper class or
5:06high netw worth income, but they're not
5:07A-list celebrities. So, just roughly
5:09like making that up, that was kind of
5:10the positioning that I wanted to
5:12represent Dr. Callaway with. Now, I
5:14actually used Claude to kind of give me
5:16some answers for what these questions
5:17would be cuz I wanted to be extremely
5:18thorough. I want you to replicate the
5:20level of detail that we have here. So,
5:21you can see the very first question is,
5:22what is the context behind the business?
5:24What do you actually sell? Who pays you?
5:26What procedures make up most of your
5:27revenue? Right? So, you can see I have
5:29that right here. board certified plastic
5:30surgeon, solo practice on Bedford Drive
5:33in LA. I do about 150 surgical cases per
5:36year. Facial only. That's very
5:38important. We don't do boob jobs. We
5:39don't do BBLs. We only do faces. Deep
5:42plane facelift and neck is the core of
5:44it. That's about 60% of the surgeries,
5:46etc. Like you can see how how far this
5:47is, right? Then we keep going. Question
5:49number two is the chair test. So if you
5:51could put one core customer, one core
5:53viewer in a chair in front of you, what
5:55specifically would be the demographic,
5:56psychographics, their beliefs? And this
5:58is really important in in forming the
5:59psychology that we'll use later around
6:01topics and hooks and scripts. So you can
6:02see here we go. Our core avatar is 52.
6:05She lives in Brentwood. Could be
6:06Calabasis. She's married. Husband and
6:08entertainment. She spent 80k to 200k on
6:12Botox filler different therapies. She's
6:14starting to realize it's not working.
6:16She wants to stay younger obviously,
6:18right? This is a classic kind of like mo
6:20for someone who's interested in plastic
6:21surgery. So that's question number two.
6:23Question number three is her dream
6:25outcome. Question number four would be
6:26her pain points. And you can see how
6:28detailed this is. This is critical to
6:30have this much depth. Again, the more
6:32you give, the better the targeting will
6:34be. And I'll show you how that ripples
6:36downstream. So, you really want to spend
6:37like 5 10 minutes answering these. Then
6:39the last couple questions are just what
6:40is your target authority statement. If
6:42you could summarize like me as a plastic
6:44surgeon, how do I want to come across in
6:46the content one sentence? It would just
6:48be this. I do facial plastic surgery
6:50explained honestly for women who've
6:51spent years on injectables and are
6:53quietly wondering if it's time for
6:54something real. As I do this, I'm like,
6:56why did I pick plastic surgery as a use
6:58case? I have no idea. You can see how
6:59crooked my face is. This is going to be
7:01fun, though. The last question is, what
7:03is your unfair advantage? Me as a
7:04plastic surgeon or you as whatever your
7:06expertise are. What is your unfair
7:07advantage here? It's one, I only do
7:09faces. Two, I have a certain type of
7:12practice in my consults, etc. Right? So,
7:14obviously, I went through this and I
7:15answered these questions, and I
7:16recommend you do the same thing in the
7:18level of depth that I have here. Now,
7:20what this skill will do before it runs
7:22anything else, we're still on the first
7:23skill, is it will take a stab at
7:25building out what your competitive
7:27position could look like in the niche.
7:28And this is before it goes out and finds
7:30competitors, analyzes videos. This is
7:33before it hits the sand castle's data.
7:34It's just guessing based on the level of
7:36depth that you provide. And so, there
7:38are these seven attributes I'm going to
7:39explain in a second for how you can
7:41carve a unique lane in any niche. So, it
7:44kind of does an initial preliminary
7:46assessment here. And that's kind of the
7:47wrap-up for skill number one, the niche
7:49interviewer. All right. Now, second, we
7:51have the unique positioning builder. And
7:53this will help you craft what I call
7:55your unique positioning angle, which is
7:57very important. This is how you craft a
7:58lane, especially if you're entering a
8:00crowded category where there's tons of
8:02people who have been crushing with
8:03content and you're just starting out.
8:04Now, I've talked about this in a couple
8:06other videos, which I will link below if
8:07you want a deeper dive, but there are
8:08basically seven core attributes that
8:11make up differentiation in a positioning
8:14strategy, right? So the first one would
8:16be topic selection. So this is where you
8:18talk about topics that are not typically
8:20covered in the niche. So if you come up
8:22with some topics and you become the
8:23first one to talk about it, you can
8:24carve a lane and have white space in
8:26that niche. The second one would be
8:27substance depth. So you're talking about
8:29common topics, but you do it in a way or
8:31a level of depth or with certain
8:33granularity and tactics that most people
8:35aren't doing. The third one would be
8:36unique stories or proof. So you've
8:38worked with certain types of clients or
8:40you've just seen more cases or you apply
8:43different frameworks that most people
8:44can't say they've done without lying. So
8:46you have an edge there if that applies
8:48to you. The fourth bucket would be
8:50avatar specificity. This is very common
8:51to carve positioning early. You are
8:53serving an underserved slice in the
8:56market. So we might say plastic surgery
8:57content is all serving women who are 25
9:00to 40 trying to do nose jobs, but it's
9:02really leaving the upper end of the
9:04market out. For example, if you started
9:05serving older women as an avatar, that
9:08could be some place that you could carve
9:09a lane. The fifth category would be
9:10delivery style. So, this is purely just
9:12how you deliver in the content. Kind of
9:14the way you do it, but also how you
9:15articulate your voice, the rhythm, the
9:18tone, things like this. The sixth one
9:19would be storytelling formats. This is
9:21classic content strategy. You're using
9:23formats that other people aren't doing.
9:24If you're the first one to kind of do
9:26listicles, breaking down the different
9:28types of surgery, but you do it in a
9:29cool way, that could be white space that
9:31you fill. And then the last one would be
9:32the visual format. So you're actually
9:34writing on a transparent overlay that's
9:36different than the classic like hype
9:38edits or different styles people do.
9:40That could be a point of differentiation
9:42as well. So your goal really in this
9:44stage and you don't have to solve it in
9:45one day is to figure out where in those
9:48buckets of seven is there a white space
9:50in the niche that your unique background
9:52and expertise can help you fill. And you
9:54can kind of take multiple of those and
9:55multiply them together. And that's how
9:57you craft your own lane to grow faster.
9:59Now, you don't have to have your own
10:00lane from day one, but the more you can
10:02carve a differentiated niche, the faster
10:04you will grow if the content is good.
10:05And so, what we're doing with this skill
10:07is we're going out and canvasing the
10:09core competitors that map closest to the
10:11topics you'd cover and the avatar you're
10:13serving. And we try to map where do they
10:16sit in those seven areas and how can we
10:18carve our own lane. So, I want to show
10:19you that right now. This first thing on
10:21the screen was just Claude's attempt at
10:23doing this without actually doing the
10:25competitor research. But now, we're
10:26going to do the competitor research. So
10:28the first thing it will do is it will
10:30build out a competitor list. We call
10:31this a watch list on Sandcastles. Now
10:34the magic is it's actually able to do
10:36this automatically if you have the
10:38Sandcastle's MCP installed. So like
10:40without you doing any work, it'll say,
10:43"Hey, these eight accounts are the
10:45closest plastic surgeons to what we
10:47think you would resonate with." And then
10:48serving the same avatar as you talking
10:50about similar things. You can see these
10:52are all facial plastic surgeons or they
10:54talk about facial rejuvenation like very
10:55closely tied to your avatar. It also
10:57recommends tier B people on the list who
10:59are plastic surgeons but maybe don't do
11:01facial stuff or they're talking about
11:03celebrities in a way that wouldn't
11:04appeal to driving leads for your
11:06business. So what Claude will do is it
11:07will recommend, hey, I think we should
11:09add all of the top eight plus number 9,
11:1210, and 12 into your list and we use
11:14that list to build the competitive set.
11:17So if you don't have any clue what your
11:19niche looks like, all you have to do is
11:20connect the sand castles. MCP and run
11:22this and it will build the list for you.
11:24It'll even create the watch list for you
11:26in Sandcastles, which I'll show you
11:28right now. So now I'm actually in the
11:29Sandcastle's web app. I've got the
11:31plastic surgery workspace that I
11:33created. And this channel list was
11:35automatically populated for me by
11:36Claude, which is pretty cool. The MCP
11:38can talk back and forth to Sand Castles.
11:40Now, I wanted to show this because let's
11:41just say Claude pulls an initial list of
11:448 to 12 accounts, but you actually know
11:46there's other plastic surgeons, like
11:48maybe there's some of your colleagues in
11:49the space, and it didn't pull those for
11:50whatever reason. There's a lot of ways
11:52that you can build this list out more
11:54directly in Sandcastle. So, if you just
11:56come into describe, you can describe
11:58plastic surgery and we'll say facial
12:01faceelifts and filler we'll say. So, you
12:04can type some keywords. We can say we
12:06only want them on Instagram and then we
12:07want them ideally between micro, small,
12:09medium. And we can search and this will
12:11pull up all of the possible accounts
12:13that we have on Sand Castles that fit
12:15this. Now, Sand Castles has hundreds of
12:17thousands of the top creators, but we
12:18don't have billions, right, that are on
12:20Instagram and Tik Tok. So, I'm going to
12:21show you how you can find others in a
12:23second, but one way of doing this would
12:24be to search for things like this, and
12:26then based on who comes up, you can see
12:28is there any white space for different
12:30people that are missing. Now, this
12:31should be pretty good because this is
12:33already pulled based on the optimization
12:35of all the answers in your niche
12:37interviewer. But just in case there are
12:38other people like Dr. Larry Fan, you can
12:40add him in here. Now, a second way to do
12:41this is let's say chestnut.md is like a
12:44perfect fit for us and he's already in
12:46the list. All we have to do is hit these
12:47double arrows and San Castles will auto
12:50resort our entire database closest to
12:52chestnut.md. So then he becomes the
12:55target instead of random keywords that
12:56you write. So this is a great way to
12:58find other people as well. Now let's say
13:00you're on Instagram and you're like
13:02there are accounts I'm not even seeing
13:03come up here. That's totally fine and
13:05expected. Like I said we have 100
13:06thousand accounts or a couple hundred
13:07thousand. There are millions or billions
13:09of accounts on Instagram and Tik Tok.
13:10So, all you have to do is go to
13:12Instagram and Tik Tok, pull the channel
13:14URL, just copy their URL, instagram.com/
13:16whatever their username is, come into ad
13:18URL, and you can actually paste it
13:19directly in here, and you can add them
13:21to the watch list. If they're not in
13:22Sandcastles, we will import them in over
13:2415 minutes, and you'll be good to go.
13:26So, what I like to do if you really
13:27aren't sure, if you know Chestnut MD is
13:30great, you just go to Chestnut MD and
13:31you open it in a new tab. And then you
13:34could also go to like who he's
13:35following. So chances are if he cares
13:37about this plastic surgery facial
13:39reconstruction niche, he's probably
13:40following a lot of people in that niche.
13:42You can actually look at who he's
13:43following. These will be sorted by like
13:44who I'm following as well. So a lot of
13:46these are not plastic surgeons. But if
13:47you go a little bit further down, you
13:49will see there's going to be a lot of
13:50people in this space that are colleagues
13:52of his. And so you can actually copy
13:53those out. Like let's just say botch TV.
13:56I could just press the botch TV. I could
13:57copy this URL. I could go to
13:59Sandcastles. I could put this in here.
14:01And I could press parse. This then shows
14:03look botch we don't have in the network.
14:05All you have to do is press plus and it
14:07will add it to the list for you. After
14:0815, 20 minutes, it'll be in there. So,
14:10the point is there's a lot of different
14:12ways that you can build out this list.
14:13But all you have to do is get this list
14:16filled out with some competitors in your
14:18niche and you'll see the rest of the
14:20engine works super well. So, let's shift
14:22back to Claude. Now, we're here with the
14:23initial list. I think it's pretty
14:24magical that you can just answer a
14:26couple questions about your niche and
14:27then Claude will not only build out the
14:28channel list for you, but also build it
14:30in Sandcastles and set up that framework
14:32for the rest of this. It's pretty
14:33amazing that it can do this
14:34automatically. And the reason why is
14:35because of that sand castles.ai MCP. So
14:38this is kind of the first part of the
14:39skill is it just actually building this
14:41foundation of the channels. So you can
14:43see I took its recommendation. I added a
14:45couple others that we had in there and
14:46we're good to go. Now the next thing
14:48that it does is very cool. It actually
14:50builds out a visualization of the
14:52positioning attributes. So if we go down
14:54and we click on this visualization, it's
14:56called the positioning wheel. So what
14:58this did is it took those seven
15:00categories that I mentioned and it
15:01mapped all of the individual accounts
15:04from our list with respect to those
15:06categories. And so it's trying to show
15:08us where is their whites space, right?
15:09So if we go to let's say topic
15:11selection, which was the first one, it
15:13says topic selection is pretty crowded.
15:14This is the full menu of different
15:15topics that people talk about. This is
15:17what we're currently seeing in the niche
15:19with respect to who's covering what
15:21topic. And here is where there's white
15:23space, right? So you can go around the
15:25wheel and all of these different
15:26positioning attributes. You can start
15:28understanding where is their whites
15:29space, what is crowded, what is not,
15:31what might be some ideas. Now you don't
15:33have to only follow this, right? This is
15:35just an initial swag at this like what
15:37does the competitive set look like. You
15:39can then start conversing with this if
15:40you want. You can like go off the path
15:42of this workflow and just start having a
15:43conversation, really asking questions
15:45and digging into this data. The only
15:47reason it's able to do this is because
15:48it went out and deep analyzed a couple
15:50of the top performers from every single
15:52one of these accounts for you and
15:53created this mapping. So, I find this to
15:55be an extremely helpful strategy
15:57exercise when it comes to what is my
15:59counter position. How am I going to
16:01position myself for a new lane in the
16:03niche, right? It also gives you like
16:05synthesis here for what that is. So,
16:07that's skill number two, actually
16:09building that competitor list and then
16:11building out this positioning wheel to
16:12understand, okay, for this person, Dr.
16:15Dr. Kevin, we click on it. Dr. Kevin has
16:17kind of differentiation in these four
16:18sectors. So now we can go open Dr.
16:20Kevin's channel, look at a couple of the
16:22videos and be like, okay, he does this
16:24format. He covers these topics. Can we
16:26be different than Dr. Kevin? Yeah, we
16:28could be different here. There's
16:29potential white space on this visual
16:30format, etc. So this isn't like an
16:33immediately tactical thing that you're
16:34going to go apply right away, but having
16:37it in your head for counter positioning,
16:38I find to be very helpful. All right.
16:40Now, the third strategy skill you're
16:41going to run, I think this is the most
16:42important one, is called the bullseye
16:44builder. And what this is doing, as you
16:46can see in the visual, is it's building
16:48out five levels of breadth with respect
16:50to the topics that you cover. A very
16:52common question I get asked, especially
16:54from business owners that are trying to
16:55make content to drive leads is how broad
16:58or narrow should I go with my content?
17:00What should my mix be? Some coaches are
17:02like, I should go super broad and just
17:04go really viral and I'll get tons of
17:05views and the followers will come in.
17:06But I never see sales when I do that.
17:08There's a good reason for that is
17:09because when you go too viral, it's
17:11actually bringing in the wrong people.
17:12However, on the flip side, if you go too
17:14narrow, every video gets a couple
17:15thousand views, you're really not
17:17trickling in sales because it's not big
17:18enough to have kind of a spike of
17:20credibility and drive enough leads for
17:22it to move the needle. So, you're not
17:24sure how you should play this. The
17:25bullseye builder is the best exercise to
17:27understand exactly what topic altitude
17:30you should be playing at. And I want to
17:31explain exactly how this works. So the
17:33first thing this skill does is it maps
17:35out in five levels what the topic
17:37breadth would look like. At the very
17:39center of the bullseye is the most
17:40narrow, right? So this would be like
17:42Elizabeth, affluent LA woman, 48 to 58.
17:45She has spent six figures on injectables
17:47and devices. She's researching a
17:48facelift. She's terrified of it looking
17:50too done. That is like bullseye
17:52bullseye. Very targeted for what we
17:54want. I would argue that's too narrow.
17:56If you just made content only for this
17:58person, it would convert at an extremely
17:59high rate, but you'd only get a few
18:00thousand views because it's way too
18:02narrow. If we go out one more ring from
18:04there, we've got filler fatigued women
18:0645 to 65. This is any woman in this age
18:09range who has been on filler, Botox or
18:11devices for years and are just curious
18:14about any type of plastic surgery. One
18:16ring from there would be any woman
18:18considering facial surgery. They haven't
18:20done filler. They haven't done Botox.
18:21They're just very lightly curious about
18:23the space. The blue number four would be
18:26any woman 35 plus who is aging face
18:28curious. They notice their face aging.
18:30They're open to anything, not just
18:31surgery, not just filler, not just
18:33pharmaceuticals. They're open to hearing
18:34about anything. And then number five
18:36would be all beauty and skincare. So
18:38we're talking about any makeup brand,
18:40any skincare treatment, right? And so
18:41you can quickly understand as a plastic
18:43surgeon, Dr. Callaway, trying to drive
18:45people to my practice, if I made videos
18:47about any type of skincare routine, that
18:49would be way too broad. It might go
18:51super viral, but like people will not
18:53come and pay me $100,000 for a facelift
18:56because of that. And so I actually
18:57really want to stay away from ring
18:59number five. I'll explain what you want
19:01to do in a second, but this exercise is
19:03extremely helpful and it generates
19:05automatically as long as you answer the
19:07niche interviewer stuff. That's why this
19:08system is so cool. You answer those
19:10questions and it kind of like trickles
19:11down, right? So, this is what it started
19:13doing is it built out an initial swag of
19:15what those five levels would be. And
19:16then it asked me to gut check like is
19:18the fifth actually too broad? Is the
19:20first the correct bullseye? Like you
19:22want to dial this in, right? So, we run
19:24this and it's correct. Now, I think this
19:26next part is actually the coolest piece
19:27of this whole flow. So the content mix
19:29that I recommend all business owners run
19:31is called the 321 content strategy. You
19:33have three buckets. Assume you make
19:35seven videos in a week. Seven videos in
19:377 days. Of those seven, you have three
19:40buckets of two and then one random one
19:42which I call the chaos agent. Of your
19:44three buckets, one of them is more
19:45broadly positioned to get more virality
19:48and followers but still acrew trust
19:49directionally in the right space. Two of
19:51those buckets, four of the seven videos
19:53are more narrowly positioned. If someone
19:55who is the right fit watched that video,
19:56they would come down the funnel. And
19:58then the last one is a chaos agent meant
20:00for random testing. So I want to explain
20:02what that means. The broad bucket should
20:04be in this light blue, the fourth ring.
20:06So it's broad enough where you could get
20:08a couple hundred thousand views. You
20:09could drive several thousand followers
20:11on one video. You could drive some
20:13conversion, but it's really meant to
20:14like stepch change your traffic to bring
20:16a lot of new eyes onto the page. The
20:19narrower buckets are meant to be either
20:20orange or yellow. So these are narrow
20:22enough like if you made a video about
20:24filler fatigued women 45 to 65 something
20:27about why filler is fatiguing the face
20:29or there's something wrong with filler
20:30those people are only one step away from
20:32a facelift plastic surgery. So they are
20:34very on target and they will probably
20:36convert at a high level. The chaos agent
20:38can be anything from red to light blue.
20:40You still don't want to go broad dark
20:42blue because if you do that and you go
20:43viral on accident, you could mess up the
20:45page. You'll bring in a lot of people
20:46that are not fits and the algorithm will
20:47get confused. So the chaos can be
20:50anything within those four. Now what's
20:52cool is this skill based on all the
20:54research it did on your competitors and
20:56the niche interview. It'll start
20:57recommending topic buckets that it
20:59thinks works well for you. So this first
21:02one will be the narrow. So this is in
21:03the yellow and orange. So it says we
21:05could do stuff about it's quily naming
21:07it the filler reckoning, but it's like
21:09what years of injectables did to your
21:10face like as a bucket of stuff. You
21:12could also do why it failed. So,
21:14anonymized revision autopsies about how
21:16facelifts and nose jobs went wrong. It
21:19also gives a couple others like it
21:20recommends those two we should do for
21:21our two buckets initially. But it could
21:23be how to vet a surgeon. It could be
21:25talking about the consult room. You have
21:27somebody in a consult, what do they ask?
21:28Why are they kind of blocked? Why won't
21:30they buy? And then some other ones,
21:31right? So, these are the narrow buckets.
21:33These would fit yellow and orange. Your
21:34job is to pick two of these buckets.
21:36Now, if you don't like any of these, you
21:37could say, "Give me more." Or, "I want
21:39to brainstorm on this. What about this?"
21:41But it's just giving you an initial swag
21:42of what might work. Now we go down and
21:44there's also some on the broad. So one
21:46of these broad in the blue, how faces
21:48actually age. So the anatomy of a face,
21:50drawing it on a mirror potentially and
21:52kind of talking about how facial age
21:54works. That's like a broader concept,
21:56but it's still directionally applying to
21:58somebody who may want plastic surgery on
22:00their face if it's the right avatar
22:02slice. Again, the surgical read, doctor
22:04verse influencer, there's all these
22:06buckets that it proactively comes up
22:07with. And this is really really targeted
22:09because it has all of the data from your
22:12competitive niche piped in. It also
22:14gives you a recommendation on these
22:16chaos ideas. I find these to be very
22:18interesting. So it says the $200,000
22:20face. What if you walked through an
22:22anonymized composite of what Elizabeth's
22:2415 years of injectables actually cost
22:26item by item and what it brought her,
22:28right? So this is talking about the
22:29money angle. That's like a random idea.
22:31Or the mirror lecture. You go up to the
22:33mirror. You have a picture of her face
22:34as a plastic surgeon and you show in
22:36long form, no cuts, exactly what you
22:38would do and like explain your thoughts.
22:40Now, these are obviously random one-off
22:41ideas that are within the red to the
22:43light blue, but they're just like
22:44one-offs. They're not they're not in the
22:46buckets. And so, what I think is really
22:47cool is that it actually proactively
22:48gives us some suggestions and then we
22:50can brainstorm from here. Now, what it
22:52does is it asks you to pick 3 2 1 just
22:54to kind of fill out the topic work
22:56moving forward. If you had to pick three
22:58buckets, one broad, two narrow, and then
23:00one chaos agent, which of these would
23:02you pick? So, it gave me a
23:03recommendation and I ended up going with
23:05that recommendation just to continue the
23:07demo. And then what it did here, which I
23:09think is really cool, is it layered on
23:11the buckets within the rings. So, this
23:15red bucket is one of our narrow ones.
23:17The orange is one of our narrow ones.
23:19And then this blue is one of the broader
23:20ones. And if you go down, this is my
23:22seven days of content right here. I have
23:23my red and my orange. I've got my blue
23:26broad around how faces actually age. And
23:28then I have my chaos agent around the
23:29mirror lecture. Now, down here are all
23:31the other ones. So, if I just want to
23:32click some of these other ones, I can
23:33switch and these will swap out, which is
23:35pretty cool. So, I'm going to keep
23:37exactly what we had as we go. But, this
23:39is like a really helpful batch one plan.
23:42Then, it says, "Hey, here's what it
23:43might look like after a couple batches
23:45of seven, right?" Cuz the whole goal is
23:46like, you post seven videos, you see
23:48what happened. If some of them worked,
23:49you double down, and so on and so on.
23:51So, it's even projecting out, hey, after
23:53two to three batches, here's how you
23:55might want to rebalance, you'll drop
23:57this bucket and replace it with this.
23:58So, I find this to be an extremely
24:00helpful visual. It also gives you a
24:01sourcing map that kind of is a more
24:03summarized view of these different
24:05levels. So, between those two, I find
24:07the bullseye builder to be extremely
24:09helpful in articulating what my content
24:11mix should look like, how broad or how
24:13narrow should I go. All right. Now, step
24:14number four is the topic scanner. And
24:16this one is very important. This will
24:18give you what you're actually looking
24:19for, which is a list of topics that are
24:22validated within your strategy that you
24:24can go make. The most valuable data you
24:26could use are the winning topics from
24:28other people in the niche, but then
24:30remix based on your own angle or
24:32substance. And the reason this is
24:34important is because certain niches have
24:35certain topic buckets that tend to do
24:37better than others. And it just goes
24:39back to the psychology of what the
24:40average viewer thinks is important. When
24:42it comes to plastic surgery facelifts,
24:44maybe they think filler is a very hot
24:46button topic that's important to them,
24:48but they think techniques for sleep
24:50aren't. So, you wouldn't want to talk
24:51about sleep if the data wasn't proving
24:53that the sleep category worked. You'd
24:55want a bias towards filler. Now, to
24:57figure this out, we're going to run the
24:58topic scanner skill, which will take the
25:00top 50 videos across all of our
25:03competitors and rank them by performance
25:06against the buckets and the avatar we're
25:08going for. Now, what's really cool here
25:09is because we've done the niche
25:10interview and we've done the channels
25:12and we now have the buckets, it's able
25:13to go through and scan a ton of videos,
25:16but rule out all the ones that aren't
25:18relevant. So, like if Dr. Mendez is in
25:20his Ferrari going up the coast on a vlog
25:22weekend with his wife. That's not
25:23relevant for me to drive customers into
25:26my practice. So, it just screens that
25:28out. It doesn't charge me a credit to
25:29run it and it doesn't waste compute to
25:30do it. So, you can see this ran through
25:32400 videos on the watch list from the
25:34last 180 days. It resorted them and
25:37screened only the top 130. 73 of them
25:40qualified. 57 of them were excluded
25:42because they were lifestyle vlogs. They
25:44were about a different part of the body
25:46or different plastic surgery procedures
25:47we didn't do, etc. And so what we have
25:49left here is the top 50 videos that
25:52apply to us. And so we run these and
25:55this takes 20 30 minutes to run because
25:57it's going to sand castles and deep
25:59analyzing. So just let this run. And
26:01what it gives you is a couple really
26:03helpful things. The first thing is this
26:04topic scanner dock. So not only do you
26:07have a performance graph on the topics
26:09like rhinoplasty, how faces are aging,
26:11etc., but you also have performance on
26:13the formats, which we'll get to in a
26:14second. Then if you switch to the
26:16bullseye placement, this shows you for
26:18every topic bucket within the top 50,
26:21where does it fall on the rings. And so
26:23you can see some of these fall in the
26:25dark blue. We want to stay away from
26:26those. Other ones are kind of in the mix
26:29in the range that we want. So this is
26:30cool just in terms of visual. But what
26:32you really are after is this top 50
26:34document. And this top 50 document
26:36really is the gold that you are wanting,
26:38right? So this is the top 50 videos from
26:40your competitors already screened with
26:42the information around the topic bucket
26:44and then the link to the individual
26:45video and the deep link into sand
26:47castles and it's all right here for you.
26:48So what you can then do is take this and
26:50go one by one and be like is this video
26:52applicable to my customer? Do I have
26:54something interesting to say? If so, you
26:56can then react and start making it in
26:58the actual video workflow that I'm going
26:59to go through. If not, you just skip it.
27:01And so as you scroll down, you can see
27:02some of these have been ruled out. Like
27:04if we go through this one right here is
27:06off niche, but most of the ones that
27:08would be rolled out have already been
27:09pre-screened. So this is like a very
27:11pure working set of outliers for you and
27:14it's organized by view and outlier
27:16score. You can even see which topic
27:17bucket each one fits into. So if we're
27:19like, oh, we want to do the filler
27:21reckoning talking about filler, there's
27:22a star here on number eight, on number
27:2410, on number 15, 16, like these are the
27:27buckets that we green lit. And so again,
27:28this is like a proactive ideiation agent
27:31working for you. This doc is extremely
27:33helpful. So, that is the fourth step,
27:35which is the topic scanner. There's one
27:37more skill in this strategy set, and
27:39then we're going to go into the actual
27:40video workflow. All right, the fifth and
27:42final skill in this strategy set is
27:43called the format finder. And I find
27:45this to be another very helpful
27:47visualization that I can actually use to
27:49decide which formats to make, where
27:51there might be white space, and where
27:53it's too crowded. Now, when I say
27:54format, formats are a little overblown
27:56as a term. It's just a combination of
27:58how you structure the information and
27:59how you visualize it. Those together
28:01make a format. People are always like,
28:02"Oh, I need to find all the hidden
28:04formats." Formats are a helpful way to
28:05structure your information, but I find
28:07it can be a bit overwhelming to try to
28:09hunt for like the optimal format.
28:11Typically, there's a couple that you
28:13like making and you just run that back
28:14and forth. Now, to help you with
28:16formats, we actually built the
28:17collections tab in Sandcastle. And so,
28:19if you go to collections and you hit
28:20this little tag and go to formats, all
28:22of the core formats that you're going to
28:24see are going to be what typically work
28:26for most people trying to acrewue
28:28expertise. And so, you can click into
28:29any of these like A versus B comparison.
28:31You can see a bunch of examples in
28:33different niches of this and you can
28:34click in, you can see the transcript,
28:36you can see a bunch of the analysis, you
28:37can actually watch the video. So, we
28:39find collections to be like the best
28:41go-to resource if you want inspiration
28:43examples across different formats. Now,
28:45in addition to collections, what's very
28:46helpful is to understand the formats
28:48that are working for your specific
28:50niche. And so, what you're looking at
28:51here is the format finder matrix. And
28:54this is pretty cool. So Claude is able
28:56to take all of those videos that were
28:58analyzed from your competitors and plot
29:00them across this matrix. So you've got
29:02on the vertical axis, these are
29:04storytelling structures. So things like
29:06case study, Q&A, problem solution, ways
29:09of kind of organizing the information.
29:11You would see all of those in the
29:12collections and sand castles. And then
29:13across the top you have the visual
29:15execution, talking head, podcast clip,
29:18the surgeon and the patient on the
29:19screen together, marking on the face,
29:21right? And so what you're seeing is
29:23where the outliers actually live and
29:25more importantly where the white space
29:27is. So you can see this visual 10 option
29:30mirror drawing. It's suggesting that's
29:32wide open. No one's done it, but it
29:34would be great given the detailed
29:35information that you shared in the niche
29:37interview. So for example, we said we're
29:39really good at going on the mirror and
29:41kind of drawing where we would do the
29:43adjustments on the facelift. So why
29:44don't we turn that into content? That's
29:45an amazing visual format that is
29:47untapped. And so this again as a
29:49resource for which formats are working,
29:51what's not working, really helpful to
29:53kind of organize your thoughts for what
29:55you might want to try. Now, to be clear,
29:57when it comes to formats, typically
29:58you're going to pick like two or three
30:00every single round of seven, and you're
30:01going to experiment. And then most
30:03people will get comfortable with one or
30:05two that they like making over and over.
30:07So, in this first phase, if you're just
30:08starting out with content, you want to
30:10be able to wildly experiment. So, I like
30:12picking ones that have either worked
30:13really well or find the white space
30:15where you think you have a unique
30:17advantage. If you're really good at
30:18drawing, maybe doing the mirror kind of
30:20overlay might work well for you where
30:22it's too hard for others to do. And so,
30:24this is the final step in the strategy
30:27flow. Basically, as a summary, what have
30:29we done? We have done the niche
30:30interview to actually give context on
30:32who we are and the avatar we're trying
30:34to reach. We've done the unique
30:35positioning builder to understand what
30:37the competitive niche looks like and
30:39where there might be white space. We've
30:40done the bullseye builder to figure out
30:42what is our content mix. What's not too
30:44broad but not too narrow that would be
30:46good at driving customers but also
30:48followers for us. Then we did the topic
30:50scanner to actually look at the top 400
30:52and ultimately top 50 videos that we
30:54deep analyzed that would fit really well
30:56within our topic buckets. And finally we
30:59went through the format finder to
31:00understand what formats are working best
31:02in our niche. If you go through these
31:04five steps, you get Claude, you
31:05integrate Sand Castle's the MCP, and you
31:07run this five skill system, you will
31:09have so much more clarity on exactly how
31:12to set up what your topic mix should be
31:14and how to actually go forward to the
31:15next piece, which is making these
31:16videos. By the way, guys, if you're
31:17liking how I'm breaking this down, all
31:19of this is available in Short Form
31:21Academy. That's my program where I work
31:22one-on-one with business owners and I
31:24really help them rev up, solve their
31:26content problems, build a content
31:28system, and then they graduate out
31:29ripping. This is exactly what I run.
31:31There are dozens and dozens of business
31:32owners in there that have had huge
31:33success. So, if you're looking to work
31:35with me one-on-one to kind of install
31:37this system and have a much more
31:39rigorous approach to winning on social
31:41media. There's a link below if you want
31:42to learn more. All right. Next, we're
31:43getting into the full tactical topic
31:46workflow and I'm going to take my hat
31:47off as the plastic surgeon. Dr. Callaway
31:49is retiring and I'm going to go back to
31:51my social media growth side. And it's
31:52just cuz I want to actually show you the
31:54process of making a tactical video. And
31:55I don't know anything about plastic
31:57surgery. So, amazing for the strategy if
31:59I'm trying to go from scratch. But if
32:00you're an actual plastic surgeon, you'll
32:02be able to give real context on these
32:04videos. So, we're shifting back to the
32:05social media growth side. I've gone
32:07ahead and I've run the topic top 50
32:10exercise for my stuff, my niche, my
32:12competitors, my channel, and I've built
32:14out that same flow here. And so, you can
32:16see I've got that topic mapping here.
32:18So, here are my 50 videos. It just did
32:20it in a different format, but you can
32:21see like all the ones with the checks is
32:22the exact same thing. Here's the bucket
32:24they fit into. For me, on my side, I do
32:26AI content workflows, hooks, and
32:27storytelling. So, it's calling out where
32:29the buckets map. It's got all the
32:31outliers. It's got the original video
32:32link and the analysis link directly to
32:34Sand Castles. Everything is the exact
32:35same. So, I ran this process to pick it
32:37up on my side. And what I ultimately
32:39want is the top 50 doc, which is right
32:42here. And so, now we're starting the
32:44process of how do we actually make a
32:45video end to end. So, once you have this
32:47strategy piece, all you have to do is
32:49start a new chat in Claude and you're
32:50going to paste in the other three
32:52skills. And these skills will help you
32:54do research on the topic, help you write
32:56hooks and scripts on the topic, and
32:57ultimately build a finished script for
33:00you, which is what you're really after.
33:02You want to understand what topics do I
33:04make and then how do I take a topic and
33:05get a script that actually works as soon
33:06as possible. So that's this part of the
33:08workflow that we're going to go through.
33:09So what I do is I look at this top 50
33:12list and I literally go down the list
33:14and I say do I have something
33:15interesting to say for this topic,
33:17something different than the outlier and
33:19would it be relevant and valuable for
33:21the person I'm trying to attract? Right?
33:23So I go down number one AI presentation
33:25generation build professional
33:26presentations in seconds with AI. That's
33:28not what I'm interested in. That's not
33:30going to drive people to SFA or
33:31Sandcastle. So I'm going to skip number
33:32two and number three were off topic. So
33:34you can see it strike it out. Number
33:36four, [music] Instagram profile
33:37optimization. Use these settings. Too
33:39transactional. I can't say anything
33:41different. I don't like it. Next. Viral
33:43video psychology. Use psychological
33:45pattern breaks to skyrocket video views.
33:47This could work, but he's probably using
33:48examples. Can I find something unique to
33:50say? I'm not sure. AI design iteration
33:52with Canva. Not a fit. Viral content
33:54psychology. Again, using pattern
33:56interrupts. I would watch this one and
33:57potentially be a fit. This one's too
33:59broad. This one's not a fit. Ultimately,
34:01I get to number 11, which is video
34:03scripting architecture. Deconstruct the
34:05ideal scripting structure for short form
34:07content. Now, that's something that I
34:08have a take on. What is the ultimate
34:10optimal storytelling structure for a
34:13script in short form? So, I see that and
34:15I watch the original video from him and
34:17I say, "Okay, do I have something
34:18different to say other than what he
34:20said? Do I have something better to say?
34:22Is what I have to say more interesting,
34:24more compelling, more succinct? Do I
34:25have some edge? If yes, I'm going to
34:28make that video." And so, that's what
34:29I'm going to do in this workflow. Now,
34:31you might say, "Wa, you just skipped
34:32through the top 10." Well, the reality
34:34is I only need one to make today and I
34:36need seven to make per week. So, if this
34:38is a top 50, I need seven from here and
34:40I can run this next week with all fresh
34:42videos and go again. So, you're allowed
34:44to pass on a bunch of stuff that's not a
34:46fit. You don't have to pick the highest
34:48ranked stuff on the top. So, I'm going
34:49to go to this chat and you can see this
34:51is what I did. All right, let's take
34:53number 11 from Real Sky Tan. This is
34:55what I believe to be true. And so, you
34:56just brain dump. I recommend using
34:58Whisper Flow or you can use this
34:59microphone and just brain dump the
35:01context that you think to be best about
35:04this idea. What do you know that most
35:06people don't know? What do you believe
35:07to be true that most people don't
35:09believe to be true? You're going to
35:10brain dump that. Now, from here, we've
35:11actually built in a research flow. So,
35:14if you want it to do additional research
35:16for you to add on to your take, it'll do
35:19that. And I'll explain how it does that.
35:20If you just want your context, you don't
35:22want any research, you just say no to
35:24the research. Right now, I went ahead
35:25and ran additional research here. And
35:27the reason why is because this research
35:28process is actually looking at other
35:31social media videos. It's not just a
35:33random 360 clawed browser research.
35:36That's not what it's doing. What it's
35:38doing is it's looking at all of my
35:39previous winners that have been deep
35:41analyzed. All of this top 50 that has
35:43been previously deep analyzed. All of
35:45the videos I've deep analyzed ever in
35:47sand castles. And then it's searching
35:48the global database of sand castles. And
35:50it's trying to find other videos that
35:52have talked about this thing perfect
35:54scripting structure. It takes all those
35:57videos and feeds into a research report
35:59for you. Now, you can choose to
36:01incorporate that stuff or not, but being
36:03able to run research in this way,
36:05essentially a super scan across all of
36:07social media is very, very helpful. So,
36:09I went ahead and ran that and I'll show
36:11you what it actually does. It ends up
36:13producing a document that looks like
36:14this. So, this is the substance sheet.
36:16So this is like what are the core claims
36:18that I said initially in the poll and
36:20then it goes down and it takes all of
36:22the research that it took from other
36:24places and combines that in. Right? So
36:26what's cool about this is what people
36:28ultimately want to do is they want to
36:29take a topic, they want to brain dump on
36:30that topic and they want the script to
36:32end up magically there. So I'm just
36:34showing you the in between step of the
36:35research. What we actually do is give
36:37you this output. So this solves three
36:38things for you. It solves the spoken
36:40hooks, it solves the text hooks, and it
36:43solves the scripts. Now, before I go
36:44through this, I just want to call out
36:46oneshotting writing with Claude is the
36:49final frontier. I do not believe Claude
36:51is able to write as good as a human
36:54today. So, the goal really with this is
36:56getting 80 to 90% of the way there and
36:58then you manually tweaking it. You tweak
37:00it, you feed in the tweaked version, it
37:02hones itself, you do that 20 or 30
37:04times, and then it starts to get close
37:06to the level that you want. On day one,
37:08this will not be able to oneshot scripts
37:10that are better than what you could
37:12write. It just can't. And so you have to
37:14understand that that's the case. Now
37:16what we've done here, which I'll show
37:17you, is we've built a datadriven engine
37:19for spoken hooks, text hooks, and
37:21scripts. And I'll explain what that
37:22looks like in a second. Let me just show
37:23you the output that you actually get
37:25from this topic. So the first thing it
37:27does is it gives you three hook options.
37:29These are the text hook options right
37:30here, kind of rendered on the phone.
37:32These are the spoken hook options, these
37:33three here. And then these are the two
37:35script options. You've got some with
37:37your context only, and then you've got
37:38some with research. Now again, you may
37:41look at these and be like, "This script
37:42is not what I would say." You have to
37:44dial this in over time. This is the
37:46first generation from this machine. But
37:47the point is, if you have a machine that
37:49you can train to get better, it will get
37:52better. And this is a datadriven machine
37:53off your own videos and the competitor
37:55videos. So what you see here is you've
37:57got the text hooks, two options right
37:58here. So the script architecture that
38:00fires dopamine twice, the 2x two script,
38:02two dopamine hits per video, how to
38:04script the real viewers can't scroll
38:05away from script structure that
38:06hypnotizes the scroll. Some of these are
38:08good, some of these are bad. I think the
38:10point is the power words are pretty
38:12compelling. So, hypnotizes the scroll is
38:14something I might actually want to use.
38:15Your hook is fine. Your lockin zone is
38:17missing. So, like the frame lockin zone,
38:19it pulled that out. That's something I
38:21might want to use. And then we go to the
38:22actual script itself. I'll turn on the
38:24research. This first one, hook, lockin
38:26zone, block one, rehook. This is
38:28modeling after the outlier. I don't
38:29actually like this frame. Here it goes.
38:31The reason people watch your hook and
38:33still scroll is because you have no lock
38:34in zone. That's actually not bad from a
38:36hook perspective to create curiosity and
38:38kind of like create a contrast against
38:40what people believe. So as you go down,
38:42this is not bad. And the important part
38:43is that it will get way better the more
38:46videos you analyze and the more you
38:48train this. So my goal is to build you a
38:50framework that you can feed in better
38:52scripts to and iteratively improve it
38:54over time. It's just never going to be
38:56able to oneshot the way you want. That's
38:58just not where AI is. It's important to
39:00accept that. Now, the last thing I
39:01wanted to show you is the actual
39:02underlying engines that are driving
39:04these recommendations, right? So, this
39:06is just one output for the topic that we
39:08decided to choose. I brain dumped a
39:10little bit, but not even that much
39:11context. And this is actually pretty
39:12good. This is a good starting point that
39:14I could iterate off of. Now, if you want
39:16to see the engine, we'll scroll down to
39:17the bottom, and these are the engines
39:19that it built. So, the first thing we'll
39:21go through is the spoken hooks engine.
39:23So, I think the the HTML visual version
39:25is better. So, what it ended up doing is
39:27it took all the videos that we've deep
39:28analyzed. It took our videos and it
39:30said, "What are the best performing
39:32spoken hooks by type, but also by
39:34psychology?" And it built this power
39:36word or power phrase cloud, which I
39:37think is pretty cool. These are all the
39:39words that seem to be popping the most
39:41in hooks that work the best. Getting
39:43addicted to your content, hypnotizing,
39:45cracked out brains, physically cannot
39:47scroll away, the opposite of slop. Like
39:49some of these obviously insane update,
39:51you're not going to use that. But what
39:52you want is just to directionally
39:54understand the words, the psychology
39:56that's actually driving people to hook.
39:58And so this is the underlying engine.
40:00Now you can set this up recurring so
40:02that it automatically adds new videos
40:03in. Every time that you write and
40:05iterate a video, it learns. This is the
40:07framework and foundation you'll start
40:08from. We did the same thing on text
40:10hooks. So you have the text hook formats
40:12and then you also have the text hook
40:14frames. And then lastly on scripts. So
40:16these are script templates. Now again,
40:18I'll say this for the third time. This
40:20is not exactly where I'd want it to be
40:22on day one. And that's because there
40:23isn't enough reinforcement learning on
40:25the writing. I want this to be tighter.
40:26I want this to be better. and faster and
40:28me to be able to write something that
40:30sounds exactly like me. The only way to
40:31do that is to give it versions of your
40:33writing that sound like you. So, if you
40:35want to run this workflow where you're
40:36like, eh, I don't even want to bother
40:38with the AI scripting. I just want to do
40:40the strategy. I want to get the topics,
40:41the top 50. I want to get that format
40:43fit. I want to be able to ramble on my
40:45topics. It give me a deep research
40:47report across all social media. And then
40:49I'm going to write my hooks and my
40:50scripts on my own. That's totally fine.
40:52I'll show you how I actually write my
40:54hooks and scripts now. And I don't use
40:56this last mile skill. I do everything up
40:59to being able to brain dump and then I
41:01write it from scratch. So this is what I
41:02actually do to write my hooks and
41:04scripts. Right? So I pull up the topic
41:05and I get that research report and I
41:07start thinking about like what could the
41:09interesting power phrases be. So I do
41:11two things. One, I go to the videos tab.
41:13I pull up my channel which is right
41:15here. I then sort by outlier score. I
41:17have all of them deep analyzed. I start
41:18with the first one that's relevant to
41:20the topic bucket that I'm making. So if
41:22I'm making an AI workflow, I will click
41:23this first one which is about AI
41:24workflow and I will read the transcript.
41:26This is how you turn Claude into a
41:27social media machine, etc., etc. Now I'm
41:29doing two things. One, if I turn on this
41:32show sections piece, I can see how the
41:34structure, the different segments or
41:36sections within the script worked and I
41:38can try to mimic that if I want or I can
41:40just look at how I'm flowing the facts
41:42through. Now, I've done this thousands
41:43of times, so like it comes second nature
41:44to me. But really, studying individual
41:46transcripts of videos you like that are
41:48mapped to a similar topic is the fastest
41:50way to get better at scripting.
41:51Obviously, I have a blank page. The
41:53first thing I need to do is hooks. And
41:54so, I'm like, what are the power words
41:56that I can use to hook someone in this
41:58case? So, I go to the hooks tab. I type
42:00in my channel, Callaway Marketing. I
42:02sort the hooks by outlier score. These
42:04are all my hooks. And I turn off this so
42:06I can see the actual words. And I
42:08literally just scan down. This is how
42:10you turn Claude into a social media
42:11machine. Here's how to plan your next 30
42:12days of content in 30 minutes.
42:14Scientifically impossible to skip.
42:15You're using the wrong story arc.
42:16Addicted to your storytelling. Like, I'm
42:18trying to look for these
42:19psychologydriven power phrases. And then
42:21I'm going to ask myself, well, what
42:23actually applies to the topic I'm
42:24talking about? So, in this case, it's a
42:26script writing structure. So, it's like,
42:28here's how to write so good it's
42:29basically unskippable. So, I could kind
42:30of adjust this like here's how to build
42:32a script so that nobody scrolls by or
42:35here's how to write a script so good it
42:37is basically impossible to skip or it is
42:40basically Impossible to ignore. I'm just
42:42like playing with these phrases and I'm
42:44writing them down. I'm reading it. I'm
42:46like, eh, too corny. Delete. H, too
42:48long-winded. Delete. H, it's the sync,
42:50but I'm not sure what it means. Delete.
42:51That is really the process of script
42:52writing. Nobody's using AI to oneshot
42:54these scripts. I wanted to show you the
42:56skills to build a hook engine and a
42:58script machine so you can improve it
42:59over time with your vernacular in your
43:01language. It's just not there today. So,
43:03this is kind of what I do on the hook
43:04front. And then on the script front,
43:06like I said, I literally just pull up
43:07the individual videos. I look at the
43:08transcripts and I just start writing. If
43:10you want to, you can copy this. Just
43:12literally press copy, put that on your
43:13page. I will never word for word take
43:15what someone's saying. But I can look at
43:17the rhythm and the flow and the way the
43:19sentences are laid out and how the
43:21points are sequentially moving and I can
43:23say, okay, they started with this block
43:25talking about the how. So what would the
43:26how be for me? Okay, I'll write the how,
43:28write a couple sentences and then move
43:30from there. That's how I actually
43:31approach script writing. It is not an
43:33exact science today. So, I wanted to
43:35show you like I use the Sand Castle's
43:36web app from hooks and scripts, but I
43:38use the workflow that we just went
43:39through for Claude for strategy and to
43:41get the topics, to get the formats, and
43:43then I'm really liking this brainstorm
43:44function where I can open a new chat,
43:46brain dump, and then it researches
43:48across social media. I actually think
43:49that's a huge hack. So, that's why I
43:51wanted to show you on the hooks and
43:52scripts. Obviously, it's not super super
43:54ideal cuz you kind of have to iterate
43:56through it. I want to give you the easy
43:58shortcut answer for everything, but
44:00that's kind of where we're at on hooks
44:01and scripts if we're being honest. on
44:02recording and editing. This is really
44:04easy. There are no clawed skills or
44:06claude workflows for recording and
44:08editing. You don't really need them.
44:09Recording, all you have to do is model
44:11after an existing video that's in the
44:13format you want to make. So, if you're
44:14trying to make patient and doctor, we'll
44:17find patient and doctor videos that have
44:18crushed. Look at how they're shooting
44:20it. What are the angles? What is the
44:22distance from the camera? How is the
44:23lighting? Who is on what side? Where is
44:26the text hook relative to their
44:27position? Just model off of that
44:29recording. What is the cadence? What is
44:30the rhythm? So that's how you would do
44:32recording. And then editing is similar.
44:33You want to take the model and then try
44:35to replicate the model timing, [music]
44:37pacing, sequencing, and camera angle as
44:39best you can. There are no great oneshot
44:42skills for this today. People can talk
44:43about how there are reotion and like
44:45different auto editing templates.
44:46They're all pretty trash. So once you
44:48have a script, I recommend recording it
44:50as close to the model as you can,
44:52editing it as close as you can. Now, if
44:53you're a business owner, you're probably
44:54not going to be doing the editing. If
44:56you are, just do it in reels or Cap Cut.
44:58Very easy. If you're not and you want to
44:59hire the guys that I work with, I've got
45:01a link below. I'll introduce you and you
45:03can get that set up. Most business
45:04owners should try to outsource the
45:05editing as soon as possible. And so
45:07that's the best way to kind of automate
45:08this. All right. Now, the last piece in
45:10this flow after you start making videos
45:11and posting them is actually very
45:13important. And this is the iterative
45:14learning loop or the system that I
45:16referred to at the beginning. Basically,
45:18once you make seven videos and you post
45:19them, you want to look at what's working
45:21and you want to do more of what's
45:22working. Well, it's kind of hard to
45:23understand why certain things are
45:25working and not if you don't know what
45:26you're looking at. And so the best way
45:27to do this is to use the Sandcastle's
45:29dashboard. So what we have in the
45:30dashboard at the top half are your
45:32aggregate metrics, which is cool to see
45:33across all of your channels. But the
45:35real power is this bottom half. So this
45:36is the content strategy audit. If you go
45:38to the reports tab, you can actually run
45:40this audit on any channel. It doesn't
45:42have to just be yours. But if you're
45:43running it on your own, this will render
45:45in the bottom of the dashboard and you
45:47can see across topics, hooks, formats,
45:50and script writing tactics what is
45:51working best for you in the last however
45:54many videos you run it for. You can go
45:56to reports and if you want to edit this,
45:58you could say, "Okay, I want to run this
45:59every 30 days or seven days, every day.
46:02I want to do it my last 30 videos, 50
46:03videos." You can configure this however
46:05you want, but once you do, it's going to
46:07show up here and give you a sense for
46:09what actually is working. So, what I
46:10like to do is I look at topics and I
46:12say, "All right, what topics are
46:13crushing?" Well, this was kind of like a
46:15one-off roundup resource library, but
46:16I've got like AI workflows, hookw
46:18writing, personal brand growth is
46:20seeming to pop, and storytelling is
46:22seeming to drop. I thought storytelling
46:23was working, but the data is showing
46:25otherwise, so maybe I should do more
46:26personal brand growth content. This
46:28dashboard is very helpful to kind of
46:29reinforce the specific things that are
46:31working and not working. Now, this is
46:33kind of just the initial swag. What's
46:35really cool about this is you can
46:36actually bring this into Claude and talk
46:38to it. And so, the way to do this, you
46:40can go to reports. You can go to any of
46:42your reports. Like this would be my
46:43Instagram one. I can click into it. And
46:45if I just click this button, continue
46:46with MCP, and I copy this in, it'll
46:49actually pull it into Claude. I want to
46:51show you what that looks like now. All
46:52right. All right. So, here we have the
46:53content coach or the channel coach
46:55skill. And really what this is doing is
46:56it's pulling in the dashboard
46:58information and it's giving me more
46:59specifics on exactly what to do. I've
47:01had it visualize some of this
47:03information in a different way here. So,
47:04I can see my follower growth over time,
47:06the topic buckets. A lot of this is the
47:07exact same from the render, but it's
47:09giving me more information. The big
47:11thing here is that I can communicate
47:12with this. I can have a conversation
47:14with my data and be like, "Hey, I was
47:15thinking about doing this other topic
47:17bucket next week. Do you think I should
47:18based on the data or not?" or here it's
47:20saying this is what I would bias for for
47:22the next seven videos and it's showing
47:23me do this more, do this less. It's very
47:26dynamic and adaptive as if you had me as
47:28a content coach coaching you every seven
47:30days. So this is what you really want to
47:32be running once you have the system up
47:34and going. It's kind of three buckets,
47:36right? It's the initial strategy that
47:37you do once every 30 days. It's the
47:40topic workflow that you do every video.
47:42And then it's the content coach that you
47:43do every 7 days. That is kind of the
47:45sequence that I recommend running. And
47:47all of this is kind of the clawed
47:49version of it. All right, guys. That is
47:51all I've got for this video. I know it
47:52was a long one. I tried to make this the
47:54master class when it comes to clawed
47:56workflows for social media. If you're a
47:58business owner, you're a marketer,
47:59you're somebody that makes content
48:00trying to drive leads to grow a
48:02business. Hopefully, you can tell this
48:03will be the most valuable channel you
48:04could ever watch for social media
48:06growth, I'm trying my absolute best to
48:08customize the videos for you. If you
48:10want to work with me one-on-one to
48:12improve your content strategy, get
48:13better at making content, but also get
48:15my eyes on every individual video, I'll
48:18review them actively and calls for you,
48:20that's what SFA is. I've got the program
48:21linked below if you want. And all of
48:23these skills with a bunch more like it
48:26are included in SFA for free. Like I
48:28mentioned, this is the exact workflow
48:29that I'm actually running on my own
48:31channels. My new marketing channel,
48:32Callaway Marketing on Instagram, went
48:34from 0 to 75,000 followers in like 4
48:36months. But most importantly, it drove
48:3830,000 email subs and over a 100,000 in
48:40attributable revenue. All because I'm
48:43running this flow, using the content
48:45coach to figure out what to double down
48:46on, and then continuing to do that every
48:48single batch. That's what you should be
48:50doing if you're a business owner, trying
48:51to use the data to get better at social
48:53media. As always, I'm trying my absolute
48:55best to push the boundaries for what
48:56content is out there. I'm trying to make
48:58every video feel like a completely free
49:00course for you. If you have any
49:01questions, please leave a comment. And I
49:03have a bunch of free stuff, all the
49:04skills in the description. So, please
49:06check those out. Let me know what you
49:07think. And we will see you guys on the
49:09next one.

About this transcript

This page contains the full transcript of How to Use Claude To Grow From 0 to 100K Followers (Full Course) by Kallaway, generated from the public captions YouTube serves with the video. The transcript has 11,863 words across 1,647 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.