How to use Jev Better than 99% of People

0:00 Jeff has just dropped, and it has made brand new use cases possible because it is 400 times cheaper than Astra and significantly faster than any other model. However, this speed and price

0:10 benefit only works if you build the apps the right way, and I'll show you how to do that with three use cases, even if you're a complete beginner. So, grab that beautiful coffee and let's begin.

0:18 One thing to bear in mind first of all, because if you don't understand this, you don't understand Jeff. Number one, we have lower cost, which means that for every pound that you spend or dollar, if

0:27 you're in the States, every item, every person, every user, it's significantly cheaper. So, the cost has gone down. And the latency. Now, the fact that it can do things so quickly makes other

0:37 economic activities commercially viable. Fancy business speak for saying, because it's so cheap and because it's so fast, things that probably didn't make sense before now do make sense. And it also

0:48 just improves the standard of most other experiences, which is incredible. Now, we're going to build this all together because this is so easy. I'm going to show you how you can build any of these

0:56 viral examples in one app. We're going to be doing this in Base 44, very generous and kind sponsor of this video. So, thank you so much for that. We're going to do all in one app. We're going

1:04 to describe it. We're going to build the app. And I'm going to show you how you can combine Jeff with other stuff to make it even more powerful. And also, what are the limitations are, which you

1:12 must, must understand. Now, this makes Jeff very powerful in my apps. So, we can get a choice, a score, or a null, and it can change that based on the input you give it. You give it English,

1:20 but it gives you something different in return. So, first I want to do, I'll put a link down below. Go and head over to Base 44. Now, what we're going to do for these app ideas is find viral trending

1:31 like apps and concepts that are blowing up on X. I found three which are cool, and I love this one here, which is when a designer gets access to Jeff. Let's say that you wanted to build this,

1:39 right? Now, this is an example. You type in something and it pops up. Maybe you want to turn that into an app. Well, we can do all of that with Base Base 44. So, what we're going to do is going to

1:47 come down and say, "Hey, I like that." You want to be someone that takes action and build this cool stuff. So, let's head over to Base 44, and I'm going to come down and drop this in. I'm going to

1:54 go go and select Fable 5.1. In Base 44, you can do auto mode. That will effectively match you with the best model for it, which is quite interesting, and this will take care of

2:04 everything in terms of the hosting. And it it's very, very cool. So, what I'm effectively going to say is, "Hey there, I want you to go ahead for me and build me an app that lets me sort two things.

2:13 One is going to be an emoji picker, so I want like a search bar and all the emojis, and I want a beautiful animation when it comes up. And the second thing is going to be everybody's worst

2:22 nightmare, a Netflix movie decider. So, what I want this to do, okay? I want to come down, all right, and I want to be able to type things in, and I want to be able to literally have a movie selected

2:33 for movie night, and we're going to call this Netflix finder. And what we're going to do is go ahead and turn this into an app. Awesome. You can do build mode, or if you want to get a little bit

2:42 more conversational, we can go ahead and do plan, but I'm going to go ahead and send this one off. And just like that, Base 44 has gone ahead and built this for us. Now, how do you actually get

2:50 this to be powered by Jev? Well, you can do that with any model that you want to inside Base 44. Also, what you can do, if you like, is if you head over to OpenRouter, you can get your OpenRouter

3:00 API key, and you can also connect that to Jev basically specifically, so that your apps in OpenRouter are powered by that. But, one of the cool things you can do in Base 44 itself is just use AI

3:11 keys that sit within it. Now, that's all managed inside Base 44 with integration credits, so you don't even need to worry about that with apps. Now, what's really interesting here, guys, is how is this

3:19 working? So, if I said something like, I don't know, uh let's go for something like diet. Okay, maybe these are food that you might eat on your diet, for example.

3:27 It pulls it up. Now, what's happened with Jev here? Now, Jev itself, every time you enter a search query, it is actually fully going through all of this information for you. And then I

3:36 mentioned it is $0 on output tokens. So, when when when he's again indexed, there is no is this good for a diet? The model is looking at the description of the emoji, and based on that, it's making a

3:47 decision. So, I could say, "Best food for a deserted island." For example, something completely random, right? You can enter that in. It tells us, I don't know, um

3:57 best food to become super fats. Okay, whatever the thing is, right? You can do whatever you want to and it will tell you. No judgement, by the way. If you like any of these emojis, I am feeling a

4:06 little hurt cuz I'm partial to a crispy chicken cheeseburger, I must admit. Now, what we're going to do is come over to Netflix. Now, guys, you know, tell your girlfriends,

4:15 your biggest problem is solved. Now, this is cool. There's actually an API you can use to connect to Netflix, which is cool, but you can see what's going on here. It is amazing at this rapid

4:24 processing. So, if I came down and said mind-bending sci-fi, as you can see, it pulls up all the different movies that we want to check out. Again, I can come down and do horror. Again, it just pulls

4:32 up any ones that I want to. Gets rid of those guys. Maybe I want to check out AI movies. And it's like it's good because it just natively pulls things up there, right? What if I said something like

4:41 Mars? And I don't know if there's any of these that are relevant to Mars. Maybe there's nothing. There's one. It's pulled up the Mars movie. Now, why is this cool? This is the big thing that no

4:49 one's talking about, guys. This is not rag. Rag being retrieval augmented generation. You know, we talk about pine cone and putting it in the big database and you have these different vectors and

4:58 we search on similarity. Jev is doing this automatically and instantaneously for us. There is no database. Jev is doing all of this. And if I want to share Netflix finder with the world,

5:07 which is what the world didn't know they needed desperately, I can just come over here and click publish, like so. Right there. Badabing badaboom. App can be public. We can run security scans in

5:16 base 44 to make extra protected. And I can click publish. I want to publish. It's all done. And now I can open it up. And we went from looking at a tweet that had hundreds of thousands of views to

5:26 having an actual app that you can share with people, share with your friends. And we can even build integrations for payments if you wanted to. And what did make me laugh, by the way, was this

5:33 tweet underneath, which is the Netflix finder, where this guy was like, I would pay for this. So, Elliot, if you're watching this, it does exist. Go ahead and and you can absolutely do that. Now,

5:42 if this is sounding like I'm speaking Spanish, I'll put a link down below for the full Agentyc Claude and ChatGPT masterclass. You'll also get direct access to my memory agentic systems.

5:51 You'll get access to my memory, my ability to talk to actual like AI itself. You get all my design systems and of course you get the full blueprints and the guide to take you

6:00 from zero to interstellar with all these AI systems. So, with that in mind, let's take a look then at the next level cuz we've done level one. What is the next use case? And again, each use case is

6:09 going to show you a different component of how you can leverage Jeff into your cool workplace. Now, what I want to do here and this is another viral one that I saw I thought was really cool is we're

6:19 going to turn creative into decisions and that's best shown by an example, right? So, let's pull up an example and walk through that together. So, this one here

6:28 effectively is scanning ads. So, the idea here is let's say you want to grow your business, ads are something that is helpful to do that, right? It's very scalable. We can go through and we can

6:37 assess it based on multiple different types and architectures and see whatever we want to do, right? Stuff that again would have been expensive and slow to do previously. Well, I'm going to come

6:46 ahead and grab this tweet as a for instance. And the way that this is going to work and this is why we're taking it to a new level here is because Jeff, one of its limitations is the fact that it

6:54 doesn't it's not multimodal in the sense that it can't see images, right? So, this is where we bring in Dr. Gemini and what Dr. Gemini's going to do in preprocessing is essentially grab those

7:04 images and turn them into descriptions and it's going to make it readable for Jeff cuz Jeff only operates with text at the moment. So, what we're going to do right here is we're going to have Jeff

7:12 to classify and score all of that evidence for us so we can actually understand it. So, we come back over to base 44 and then drop in a tweet. I'm going to say, "Hey there, I would like

7:20 to go ahead and build for me an ad app and this ad app I'd like to be inspired by this design. But let's take it to a new level and do some different things. I want to be able

7:30 to give it categories of anything that I want to rank. I want to be able to select individuals that are perhaps competitors to me and I would love to have this all indexable by Jeff such

7:41 that I can ask it questions and it can find the right ads for me based on what it is that I'm actually trying to achieve. And then just like that guys, we have something from Base 44. And the

7:49 cool thing is when Base 44 is building again, you can change the models. It will ask you questions to help you nail out. Now, what I wanted to really do here is get under the skin of what Jev

7:58 actually unlocks with this because it's really important to understand. So, check this out. What's going on here? We've classified hundreds of creatives against a custom rubric. So, hook, type,

8:07 angle, clarity, landing page mass mass and one gun. Secondly, the cost is what makes this whole library tractable. This is the cost. There are so many zeros, I wouldn't have enough time to actually

8:18 read them all out. Per classified ad, processing 400 ads cost cents over $30 for Frontier Model to do that. So, we've got over 400 ads here. We've got some examples with Hormozi and School. And

8:29 again, that would cost us a lot more money to do that with Big Ad. Now, the fact that it's fast means it's a live workspace. It's sub-second per ad classification. Means a strategist

8:38 watches the grid fill in real time and can rerun with a new rubric instantly. So, as you have these thoughts, you can just pull that ad up straight away. Like, what about if we did this angle?

8:48 What if we did that angle? Jev gives us the speed that pays the bills. So, essentially, the speed and the cost change the unit economics. So, we've got all this stuff here. Awesome. So, let's

8:56 come down and do one of the search libraries. So, this has got a load of stuff on I did Alex Hormozi and School and a few other people. Has hundreds of ads. Let's ask it a question. How about

9:05 Let's say we're thinking, let's find a UGC style creatives. I asked that as a question. Bam, we now have three UGC style creative. Sign up UGC style. Let's check this one out right here. As you

9:15 can see, that's handy. Which ads lead with a discount? Let's ask Jev that. Bam. Again, stop hoarding free PDFs. $9 a month is enough. We click on this, we

9:26 can have a look. It's offer led, creative format, free trial, free trial period. Interesting. And we can do something else. Let's ask show whose ads whose landing page doesn't match. I

9:37 guess that's an interesting one. Let's see what that is. But you get the idea. You can essentially ask it any question that you want to. And what's really cool is it's been scored on all of these

9:45 different levels. And here's the evidence. Here's all the tags. It's incredibly helpful. So with this, we can build our ad libraries from our competitors. We can make it super

9:53 up-to-date. Had scraping tools into base 44 with things like Appify. And effectively, we can just search for the things that we want to based on what you want to do. It could be hooks. It could

10:03 be hooks outside. It could be anything that you like to. But that leads us on nicely to the third level. And then I'm going to tell you some limitations you need to know about. But level three is

10:11 very powerful, guys. Level three is an interesting one. And for for the designers out there, I know you're watching. And I love that you're watching. Describe it and find the

10:19 component. So what we can do with Jev, again, remember, this is so important to understand. When it comes to Jev, what is it doing? Well, basically, it can choose. You can give it several

10:30 options to make a choice. It can give you a rank, right? And it can also tell you an answer to something. So it's very, very powerful like that. And when you think about it from that

10:38 perspective, you can get Jev to do many things. So why don't What if we had a huge live, you know, component libraries? So for example, let's say that we had 21st.dev. 21st.dev has all

10:48 these different components, right? Look at these components. It's crazy. It's ridiculous. What if I could actually index all of these things? Or not even index them. What if I could just attach

10:57 Jev to this site? And then based on any query I had, it could find all of them. Now what you'd need to do without Jev is basically index everything. And there are thousands and thousands of these

11:07 things. Jev means you don't need to do that at all. And everything is instantly indexable. So we can give it, okay, a brief and library. Jev is going to rank it. And then we're going to get the

11:16 right starting point to build anything. Maybe we can build our own app, build our own design builder, right? We could say, "Hey, I want a sidebar." And I could say, "Jev, based on this website,

11:24 find me the best sidebar." And within seconds, it finds that sidebar. And then also the same thing, I need a new login screen. Jev, what's the best login screen based on this app? Jev will go

11:33 and find it for us, which is incredible. And again, this would be so expensive otherwise. So, let's head back over to Base 44. I'm going to say, "Hey there, I'd like to go ahead and build for me an

11:40 app. Essentially, I want a UI compo- component searcher. Go and grab some interesting assets from online. I want the ability for Jeff to search through those and find me the best components

11:50 and assets based on my search query." And then Base 44 works its magic and brings it back. And I wanted it to get it to explain what this app is actually done for us so you can understand where

11:58 Jeff is adding the value here. It ranks 255 components that it built against a plain English brief. You just yapped to it and it does it, okay? Every candidate gets

12:07 scored individually. It's not retrieved based on keyword. And I can tell you, I've spoken to billion-dollar companies, guys, that used to do keyword research and it leads to all kinds of unknown

12:16 levels of chaos. And honestly, when you're doing things like this, and again, maybe you want to do UI components. Maybe you don't. The point is, at some

12:23 point, when you're doing anything of scale, you need to index stuff, right? And you will run into the problem of, how do we describe the thing that we have? How do we actually do that?

12:32 If it's images, we can use things like Gemini to kind of articulate those. Now, what's interesting here is we can ask three questions per component, right? Requirements, function, visual fit. So,

12:42 around 750 model calls per search, and it returns it in a second or two. Like, that's the kind of thing you're getting here. But what's really interesting is that it's judging instead of matching a

12:51 pricing table with a monthly annual toggle. It hits the toggle requirement, not the word pricing. So, it just adds a lot of additional value here. Let's give this little play. We've got everything.

13:00 We've got blogs. And guys, it you know, Base 44 just spun this up with with Fable 5.1 chipping in for some help and making some magic happen. Again, we can pick any model we want to. What if I go

13:10 down for something like login? Okay, I want to login. Let's have a look at this. Cool. Let's see what we can get out from a login point of view. And just like

13:16 that, 11 strong, 12 shown. Again, cuz we get rankings, which is cool. But then let's get specific. Maybe I want like a purple one or something that's like animated. Animated light beam login,

13:27 okay? I'm just describing what we've got here. Let's see what pops up for this one as we go down and find it and bam, there you go. Animated light beam login it has picked up and it's just indexed

13:36 the entire thing instantaneously for us. And then if you wanted to, this is a really cool idea you have is you can magic build these things. So if we say AI note-taking launch page, again you

13:44 can give it commands and just have it build stuff. Obviously, you can do this in the app but you get the idea of what you can do now with Jev on this platform. Now, mastering Jev is one

13:51 thing but if you don't know how to use the best models to drive maximum value in your life and your business, you're just leaving too much on the table. So we're going to learn how to do that

14:00 exactly together in this video right here.