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.
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.

Jack Roberts · @Itssssss_Jack
This video has no Most replayed graph yet: YouTube shows one only once a video has enough views. These are the moments viewers replayed most in Jack Roberts's most watched videos.
Most replayed moment at 7:07
3.9x that video's typical replay level
can download all of them. And then, what we can do is then integrate these icons into any kind of animation or graphic. And this is an example. I just gave it the packs and said, "Hey, design me something interesting." Now, check this out. This is just something on travel. So, this didn't have to be a trip. And you
Said at 6:59
The graph counts replays. It does not show where viewers stopped watching.
Words
3,000
Runtime
12:16
Speaking pace
245wpm
Reading time
13min
245 words per minute, above the 201 75th percentile of 349 measured videos. That distribution comes from the 349-video hook study.
Opening (first 30 seconds)
Jiff is a new model but it's incredibly fast and very cheap. And in this video I'm going to show you some incredible use cases. I'm going to show you how effective this thing actually is. Look at how it compares to Claude. And most importantly, I'm going to show you how you can actually apply these use cases that you may not have been aware of. I'm going to show you how you can use this to go ahead and build beautiful reels. I'm going to show you how you can use it in your agentic operating system. We're going to build a lot of incredible stuff with this and check out the new capabilities. Hey there, Jarvis, could you do me
123 words, the words spoken in the first 30 seconds at 245 words per minute.
Free, no signup. See how the first 30 seconds hold attention, with rewrites.
Sentence shape
| Measure | This transcript |
|---|---|
| Sentences | 236 |
| Average words per sentence | 12.7 |
| Longest sentence | 52 words |
| Questions asked | 40 |
| Sentences containing a number | 17 |
Most used terms
Filler phrases
74 in total: like 17 · actually 10 · um 9 · right? 8 · kind of 7 · literally 6 · you know 6 · sort of 4 · uh 4 · basically 3.
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.
Free, no account. See where attention is likely to drop, with a rewrite for each weak line. The free check shows the scores and the one issue costing the most. Or run it on the words above first.
Free · No login · See a sample audit first if you prefer.
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.
Jiff is a new model but it's incredibly fast and very cheap. And in this video I'm going to show you some incredible use cases. I'm going to show you how effective this thing actually is. Look at how it compares to Claude. And most importantly, I'm going to show you how you can actually apply these use cases that you may not have been aware of. I'm going to show you how you can use this to go ahead and build beautiful reels.
I'm going to show you how you can use it in your agentic operating system. We're going to build a lot of incredible stuff with this and check out the new capabilities. Hey there, Jarvis, could you do me a favor? Could we just head over to my memory system, please? >> Memory's open. Don't strain yourself. Take a look around. >> Excellent, Jarvis. I would like to start a bit of a research session now. Basically, create a new session with the model and just do some research on Glider for me. >> Research session spinning up in a separate task.
Feel free to peek at the chat for updates. Off we go. >> Excellent, guys. And now what's really interesting here is it's used Jiff and this is just one of the examples that I'm going to show you. I'm not even going to show you what this agentic operating system can do. I'm also going to show you how you can actually build this yourself and how you go through this entire process. Now look, it used Jiff very quickly to route this dynamically to a specific model.
I'm going to cover all of this inside this video. Now so grab that coffee and let's dive straight into this beautiful system. So, first thing to understand with this whole agentic operating system and Jiff and Claude is the fact that it is just way faster. It is 40% cheaper than Astra, significantly faster, and it is, as you can see, in many respects, better. It can generate comments. It can identify things for Instagram reels.
For example, I've been using it to market myself. It can help you pick models, send emails out, do taste checks, landing page tasks, you name it. And we're going to bring this together inside the agentic operating system. We're going to show you how you can decide. We're going to cover how you can use agentic voice with Jiff, how it can go ahead and pick incredible models at lightning fast speeds like you just saw. You're going to see how it can actually help you with reels using Opus 5.5's new incredible technology to build beautiful graphics, how it can sort your inbox, and lots more.
So, let's begin now with the first use case that I really want to dive into. This is really, really powerful here. Is essentially, um how Jev itself can go ahead and make decisions. So, if you look at this, for example, hey, uh let's say you ask Jev a question, right? Um how do I run my agency? Where do I sign up for this course, right? And it's trying to identify, you know, who is this individual? Are they, for example, a buyer?
Is it a question? Is it praise? Is it criticism? Or is it spam? Now, let's take, for example, if we compare this with Opus 5.5. Bro, this changed how I run my agency. Where do I sign up for the course? Jev can say straight away, this is a buyer, and it does that in less than 1 second. Whereas Claude Opus 5.5 took a little bit more time to do that. So, let's go over now to the Ajanctic Operating System and check this out.
And of course, there's many different things. It doesn't just look good. It has a high levels of performance with this. So, let's go now over to our inbox, as a for instance. And as you can see down here, I have some sample emails. I put them as demo, because I'm demoing a YouTube video. Now, look at this. Let's say that we want to sort and categorize with Jev. Well, check out what we can do. If I click on sort with Jev, for instance, you can see this.
Look at the speed of that. Look how fast that was. And now it's uh sorted everything into replies, newsletters, billing, community leads, sponsors, and FYI. And it did that in real time. So, you can see just how powerful this thing is. My next incredible thing that Jev can do is dynamically route all of our conversations through to the right individual. And it can do all this using a voice operated system, which is very cool.
I've got loads of examples on here that you can play around with. So, let's come back over to the Ajanctic Operating System now. Check this out in action. Hey there, I would just want to understand how much faster is Jev on average than Opus 5.5, for example? >> Oh, look who's getting geeky now. Jev's speed is typically about 20% faster in prompt execution than Opus 5.5, depending on the task. >> You're getting pretty cheeky, I must admit, Javis. >> Just keeping you on your toes.
So, anything else you'd like to test? >> So, why is that interesting? That's interesting because what's going on underneath the hood here with Jev is it is dynamically making decision. Should the query go through to its own simple website capabilities or should it for example delegate it to Kodak? Should it delegate it to Claude? And also it's deciding behind them the scenes here which model it should go to. For example, you saw here it decided that Opus 5.5 was the best option.
Cuz look, it has five options. One is a quick answer, the other one is open up a page, show a memory, or Claude Opus 5.5. And you can see that breakdown here. And if you're wondering about how this voice by the way you can choose any of these different voices and systems very easily here. >> Take a breath. You've got this. Let's start with the first thing. >> Okay, but obviously the best one is Jarvis, right? You can pick the speed and also it's humor mode if you want to.
You can also give it like system prompts. The way that I've done this by the way is I've integrated with something called fish audio. I'll put a link down below so you can go ahead and grab this. This is ridiculous. All you literally need to do guys for this is come over to developer, literally create an API key, and you can connect it. It's pretty outrageous, I must admit. I've been having a lot of fun with this. They've got a full library of different voices that you can grab.
And the reason I mentioned that out guys is not only because does it enable you to have these conversations but check this out. It is three times faster than typing, and we also want it to be a little bit fun and have some great stuff. So essentially what it is it's a lot cheaper than other models you'll get, which is why I want to call it out. It's really freaking sick. Um cheaper than some of the comparable out there.
So you might find yourself saving a few dollars. Um it also has sound effects, which I'm going to show you when we come to do the Instagram Reels section with Jeff. And also you got one API that can do everything from cloning your voice, uh doing that for Instagram Reels in one simple system. Um again, you can actually have a conversation with it and and literally link up any voice that you want to. So if I'm over here and I'm wanting to understand this is even more important.
I get to text a speech. And on this section here guys, I can come over and pick the voice, right? I've got the model 2.1 is the best one. I can come down and pick the speaker, go ahead and choose anything I want to, add a new voice. And here I can go ahead and find them, get loads of stuff. And then you can integrate any voice that you want to in your OS, which is exactly how I've got this kind of Jarvis voice in here.
Now, the next level is a really interesting one, which is Jeb's ability to actually edit. Now, in my last video, I showed you how effective Opus 5.5 is at getting this stuff. So, what it can do is make a lot of many decisions for us. So, if I head over down to our Jentik operating system, let's go over and check. I find this how really helpful, by the way. So, let me just bring up um Jarvis here real quick. Hey there.
Quick question for you, dude. How much um usage have I got left on Claude right now? >> You have around 57% of your 5-hour total running left, and around 83% of your week total. >> Okay. Okay, let me just butt in right there. I I do me take me over to the design section of the OS. >> Design section is on. Try not to redecorate your universe all at once. >> Appreciate that. Now, I'm going to come over here and just shut him down, cuz he needs to relax a little bit.
Now, what you can build in this, essentially with this, is anything you want to. So, one of the things that I'm focusing on right now is short-form content, right? So, if I can show you, I come over here to YouTube. This is a short that I did, and I generated a lot of graphics here with um basically Opus 5.5. So, if you check this out, you'll see. You can see it's got these graphics. These graphics here were generated by Opus 5.5.
And even this one here, for example, is really cool. And you can see how it how it basically goes ahead and does it. So, the idea with this, what we can do in our Jentik operating system, guys, is effectively drop them in. So, if I come down here to Let's click on a new reel. I can come down, choose a video, and essentially decide what do I want to do. Do I want to clean up the audio? For example, less noise and clearer speech.
Again, we can use all this sort of stuff. We can cut the pauses out, and we can also add sound effects by using Fish Audio. We can drop in any reel that we want to. And when that's done, it goes ahead and produces it. So, if I check out an example I did before, you can see what it does is breaks down the entire recording that you give it, and it did all a lot of this editing, which I thought was incredible. It puts you in different places, adds in ti- um kind of like subtitles that you can play around with, which is really cool.
And effectively, what we can do is set up certain requirements about what assets should it build. So, I can come down and say, "Hey, let Jev pick which is the best one to do this." And you can add in, as you can see, different requirements that you think are good, and Jev can actually come down and also select for you based on everything that you're doing. So, it can pick out what it perceives to be the best one and why, like energy, 1 second read, fresh look.
Is it good or you know you may say, "Hey, what's really important for me is brand consistency. And does it have Is it got visual hierarchy? Has it got XYZ?" And you can kind of pick this out yourself and leverage Jev to do this kind of thing for you. And then once you've got this over here, you can just export it, milk it together one clip. We've been having a great time with this sort of stuff on YouTube. And I'll be honest, guys, I'm super impressed with the kind of thing you can get.
Let me even show you the real that I just a second publish, for example. You can see how we're leveraging it in here. It's in time. You can see down some of this stuff is literally created by Opus 5.5, and some of it is different assets from different locations. But I'm genuinely massively impressed with the ability of Opus to do this kind of thing. And again, we can use Jev to help us make better decisions with this in many different ways, including getting sound effects that we can directly integrate because Fish audio itself has sound effects over here that you can just grab and use.
And if this all sounds like I'm speaking Spanish and you want to get this entire engine to operating system, it's a one-click download, and learn how to build beautiful interactive systems and all this sort of stuff, put a link down below for my full Clutter Code Masterclass inside my community. I'm literally updating it this week with an incredible new design system. You don't want to miss that. If you want to get laser ahead and grab this beautiful OS for yourself, I'll put a link down below so you can grab it.
Now, the other thing to really sort of bear in mind with this is Jev can do lots of other things. Like what? It can, for example, check invoices about what needs to be paid. So, let's take a look at this one, for example, right? This is not going through all these different invoices, and it'll miss essentially like just a few seconds to identify what needs to be paid, what we need to check, and it did that ourselves.
So granted, we can check all of them, and you can see it working through now, whether it needs to be paid, pay, file, check. So you can essentially do this at any point you want to. Where does it sit in an actual process? You get an invoice, your system adds to what it knows, then Jev checks it, and then you can actually use this as a filtering mechanism to make incredible choices. You can even come down and look at things like this, then.
New bank details paid today. And it can add a riskiness factor. Cuz it's so cheap, guys. Like it's literally $0 output tokens, and its input tokens are unbelievably low. Like it's essentially you may as well add this as an actual classifying model to all of your systems. It's very, very cool. The other thing we can do is search by meaning. You know, months ago, you'll know that I had over here, guys, if I show you this design operating system in the OS.
If I come through to my library, then I can actually search by meaning. Again, and I'm using lots of assistance for this. I can come down and search for burger, as I often show, and burger appears. Now, Jev itself, guys, can't actually read images. It can't. So what you'd need to do in this case, and I've built this into my Agentic operating system, is the ability to essentially have a very cheap model that kind of looks through all the images on your entire computer and gives it a description.
With that description, effectively what you can then do is have everything there. So it's now got, you know, let's say this burger here, for example, doesn't have the burger anywhere in the meta description. No freaking problem. Now we've assigned that with the cheap image model, Jev itself can easily do this filtering and finding unbelievably quick. Again, you can ask Jev and search and do whatever you want to. Jev is ranking through the images, it can go through and find anything for you.
And it doesn't just stop with image search, guys. We can do things like searching by meaning. We can go ahead and get a design taste check on everything you want to. So say, for example, you give it a generic AI page. You can say, "Hey, is this page generic?" Yes or no. Uh you can give it this one here. "Is this one on the Obviously, these are just like general examples, but you get the idea here, right? Anything you build, you can run it all through Jev yourself.
Because essentially, the way that Jev is working is only giving you three things. Number one, it's going to tell you are they a It will give you like a Is it A or B? So, it will pick one. Is it a buyer? Is it not? It will give you a yes no example. Like, "Hey, is this text a scam?" And it can also give you a score. We can use this effectively for whatever we want to save loads of time, right? You can ask it things like, "Is this review fake?" Yes or no.
"Is this lead hot?" Cold or not. "Is this text a scam?" "Is this safe to post?" You know, "Is this a business expense and which team should take this ticket?" And it can go down and just give you that information. So, the next question becomes, what job should you be giving to Jev right now? But you can only really know that if you know how to apply Jev across your entire business. And we're going to cover exactly how to do that together inside this video right here.
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: paste a draft and see where it stands before you record it.
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.