Getting the transcript
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.
Reading the captions from YouTube. A video nobody has opened here before takes 10 to 30 seconds; this page fills in on its own.

Omar · @lighttrok
Words
2,211
Runtime
11:07
Speaking pace
199wpm
Reading time
9min
199 words per minute, between the 181 median and the 201 75th percentile of 349 measured videos. That distribution comes from the 349-video hook study.
Opening (first 30 seconds)
Hey guys, it's a beautiful Monday. And in today's video, I'm going to show you guys how to properly train your Claude. Meaning, when you leave this video, you're going to understand exactly how to use Claude to your advantage by making him work for you instead of you sitting and typing on your keyboard, prompting, and do all of that boring stuff. So, without furthermore, let's get started directly. I have an example for you guys. This is an example product. It's an OSHA product. I'm telling you guys, I'm not running this and you shouldn't run this because
100 words, the words spoken in the first 30 seconds at 199 words per minute.
Free, no signup. See how the first 30 seconds hold attention, with rewrites.
Sentence shape
| Measure | This transcript |
|---|---|
| Sentences | 145 |
| Average words per sentence | 15.2 |
| Longest sentence | 59 words |
| Questions asked | 6 |
What this transcript is
Every word below is the caption track YouTube publishes for this video, pulled from the video itself and reproduced unchanged. It is not Prepublish's writing, not a summary, and not a re-transcription: it is the video's own published captions. English captions, generated automatically by YouTube, in the video’s original language. Source: the video on YouTube. A channel that would rather this page did not exist can ask for its removal through the contact page, and it is removed.
Hey guys, it's a beautiful Monday. And in today's video, I'm going to show you guys how to properly train your Claude. Meaning, when you leave this video, you're going to understand exactly how to use Claude to your advantage by making him work for you instead of you sitting and typing on your keyboard, prompting, and do all of that boring stuff. So, without furthermore, let's get started directly. I have an example for you guys.
This is an example product. It's an OSHA product. I'm telling you guys, I'm not running this and you shouldn't run this because it doesn't exist. So, first thing, you find the video, concept, or thing you want to be saved in Claude so you can reuse it for the future. So, this is going to be the example video for me. I just take the link. I've already taken it. Bam, then download it. And the next step, we open Claude.
So, first things first, we're going to take the first frame and end frame of each clip in the video or the thing you want to copy. It doesn't have to be a video, it could be simply a prompt, it could be something you are specific about. Maybe you say three-to-one concept and Claude understands to do the three-to-one concept directly because it has everything stored. So, the first thing for me, I'm going to copy the first frames and the end frames and come back to you guys.
So, here we go, we're done. All first frames and end frames all in the chat here. And to keep it very simple for the sake of this video, I'm not going to generate any video or image yet. I'm going to use this as an example. Let's say you have your first frames and end frames already ready or you have something you want to save already ready. What are we going to do to make sure Claude properly understands to reuse it in the future?
Let's say I want to save this hook here. This is the first frame for it and this is the end frame. What do I tell Claude? I tell it the following. Listen closely. The first attached image is the first frame and the second attached image is the end frame. I want you to save this and I want you to save the prompt I'm going to attach below and this hook is going to be called the shocked coffee drop panning to the right hook and we are going to reference it as the store concept.
Every time I tell you store concept, it means you should use this first frame and end frame, but if I have any specification to want to change the avatar, you change it to the avatar I attached in the future and use the same prompt and same exact store as we see here in the chat I attached. And bam, guys, this is how simple truly is. However, I wasn't really specific here. When you do this, be more specific. Tell it exactly how the background looks like.
You don't have to tell it if your Claude is already trained, but I'm assuming for most of y'all, the Claude isn't truly 100% trained. So, you need to be very specific and make it understand what's to keep, what to not keep, what to change. Maybe in the future, you want to change the avatars in this first frame image, so you just attach an avatar you want to replace these two guys with and Claude is going to understand directly what you mean by attaching two avatars because I explained it exactly here.
When I attach another avatar, Claude understands that it should replace my new avatar with these two guys or maybe one of them. You tell it keep this guy on the left and replace the guy on the right with the avatar I attached in the chat. And the most crucial part is that you reference it. Meaning, you need to name it something. If you want to do the three-to-one concept and you want to save it, you have the prompt, you have the first frame, end frame and the prompt below, you tell it this is called the three-to-one concept.
Analyze it and make sure you save it in memory as the three-to-one concept and every time I tell you do the three-to-one concept, you should do it with the prompt I attached to you. There's two ways to go about this. You either do it the method I've shown you here, which is the best method, or you can do it the skill way. Meaning, you just save it as a skill instead of referencing it like we did here. For the chats, it does remember from memory.
However, it's better to attach the skill always when you want to use it. But, the reason why I don't use it is because you need to always click here and choose your skill. But, with this method, I can just say the word. It pretty much works the same. But, I prefer this method of referencing what I want, 3 2 1 concept, store concept, and then Claude does it directly because it's saved in the memory as the 3 2 1 concept.
And if you prefer to do a skill, what you can do is just do exactly as I did here, and just tell Claude I want to create a skill out of the things I wrote here. And it's going to create an MD file, a skill file, and it's going to save it automatically if you, of course, tell it. Tell it to save the skill after it is done. And it's going to save it. You can reuse it several times just through the chat here. Just to show you guys how it works, I'm going to demonstrate exactly how to do it.
Here, I have the prompt. As we see, I pasted it. This is a prompt I made, and it's basically for this concept. And what I'm going to say is, save this prompt as the store shocked concept. And the first attached image is the first frame. That's how it should look like, two men holding coffees and looking at same direction, doing the same exact thing as in the image directly. And for the second image, that is how it should look like.
And if I want to change the product box to my product box, I'll tell you, but keep it exactly the same. Now, it's loading and it's saving everything I've told it, the prompt, the first frame, the end frame, and the things I've specified here exactly. So, when I say a certain word, it's going to do exactly what I asked for up here. Okay, now it's done. I'm going to tell it, fire me a video of the store shocked concept without any first frame or end frame, just using the prompt.
Bam, we send it. If you guys are finding these videos valuable, make sure you drop in the comments what you want to see next, something you want me to go over specifically with AI, or maybe just organic drop shipping in general. Bare in mind now, the video is not going to turn out the best because I don't have any first frame and end frame. Because the first frame determines how the video is going to turn out. The prompt itself, most of the time, won't get the job done.
So, you need a first frame and an end frame is something just extra if you want to show something else towards the end of the video. But, if you only have like somebody shocked being shocked like this and not showing anything else, no shelves, no nothing in the store, then you can just attach a first frame and that's enough. All right, the video is done. Let's watch it. >> Oh, oh my lord. >> You guys see? That is how simple it truly is.
We didn't even have a first frame. And look how similar it is to the original image. Imagine if you had a first frame. And that one you can also save it. And I need you guys to understand that Claude specifically generated this video in Settings 2.0, 9:16 ratio, of course, and with 4 seconds specifically and with the audio on. You know why? Because I fed it with the information of how I want it. Do you see here? It's specifying that it run it in Settings 2.0, 4 seconds, sound on, 9:16 ratio.
That's for a reason. It's because I prompted it. I told it if a video have some sort of speech, some sort of shock, it should always create it with Settings 2.0. Because with Settings 2.0, it's going to turn out the most natural and most realistic. I told it to use Kling 3.0 to save on the credits. Because it's not going to be noticeable if only a girl is shocked because there's no much movement as in this video. There's movement here.
There is they're moving, the drink is dropping, then there's a pan towards the right. So, Sedins can handle this much better than Kling 3.0 does. I'm going to show you guys how I did it. So, it uses Sedins 2.0 and there's movement or anything complex going on in the hook. Take inspiration of what I'm going to tell it now. You can change anything. You can literally tell it to use Sedins 2.5. You can tell it to use maybe Ben Nano Banana instead of GPT-2.
You can prompt it exactly how you want it. And this is a one-time thing. You tell it and it lasts forever. From now on, I want you to use Sedins 2.0 when there is movement in a video or there's any speech or anything complex. So, use Sedins 2.0 every single time there's movement or anything crazy going on in the hook. And all videos you generate by default should be 5 seconds duration. And all of the things I've told you above applies all of the time unless I tell you otherwise.
This is the rules. And if I tell you, for example, to make the video 3 seconds, then you make it, but only specifically for the video I've told you. You can go even deeper and tell it whatever you want. Literally, if you want to save something, you tell it and you reference it or you just tell it like this and it lasts forever just because I've told it. These rules always apply if I don't tell anything otherwise. So, these rules that the video should be 5 seconds always and Sedins 2.0 should be used when there is anything complex going on in the video like talking, walking, dropping a coffee.
It should always be used. And if I tell it otherwise, maybe I want to use Kling 3.0 this time. I tell it specifically, "For this video only, I'm going to use Kling 3.0 instead of Sedins 2.0." With that being said, guys, this is how exactly you're going to save things properly inside your Claude chat. And every time you use this, remember that you are saving time. Because think about the person who writes everything manually.
He brings a prompt, a new prompt, and he pastes it, and he brings a new first frame, a new end frame, and then he copies this video. It's too bloody slow. If you compare the guy who uses this method, which is referencing what to use just by typing two words, one word, you choose. And Claude does the work for you instead of you having to go back and forth, get a prompt, get a new first frame, get a new end frame. It is too slow.
So, you need to pick up your game and actually use AI stuff to your advantage. And these methods will save 10x of your time. I guarantee you, if you use this, you're going to see how much of a speed you have when you push out content. And as we know, if you don't post content, there's no money. No money if you don't post content because people can't see your vision. They can't see what you're selling if you don't post it out there.
So, what do we do? We leverage the things I've told you in this video to make the most out of Claude. With that being said, make sure you join the Discord in the description so you can stay up to date with the market all throughout the Q4 and get daily test orders for free. And if you have any other specific question, make sure you DM me on Instagram. I'm looking for some creators. If you have made money with AI and you have a lot of experience or some experience, you're intermediate with AI, make sure you DM me on Instagram, creator, and I'll get back to you, and we'll run it up this entire Q4.
Hopefully, this made sense, and hopefully, it helps.
The words are the caption track's own and nothing is reworded or re-transcribed. Paragraph breaks are placed between sentences so the text reads as prose.
Free tools for your own script. No signup, no login.
Paste your draft and see where viewers are likely to drop off, with a rewrite for each weak line.
Paste the first 30 seconds of your own draft for a hook score and rewrites.
Check your draft against YouTube's advertiser-friendly guidelines before you record it.
Read this channel's public videos and transcripts, and download a writing brief for it.
| Sentences containing a number | 21 |
Most used terms
Filler phrases
15 in total: like 8 · literally 2 · sort of 2 · actually 1 · basically 1 · you know 1.
A literal whole-word count of the same phrase list the Prepublish browser extension uses, so a phrase inside another word is not counted and a phrase used in its ordinary sense still is. It is a count and not a judgement.