Unleashing the Power of Claude Sonnet 4.5

0:00Hey guys, Lady Luck here. Are you going on any road trips this summer? I know I'm going to be going on a bunch of road trips. And being that I'm going to be Passenger Princess, I love playing on SpinQuest.com. SpinQuest has all of my favorite slot games, live blackjack, live craps. Head on over to SpinQuest right now and get yourself a$30 coin pack for just$10. SpinQuest is a free-to-play social casino. Avoid where prohibited. Visit SpinQuest.com for more details. You're listening to a podcast right now, driving, working out, walking the dog. If you're into podcasts, chances are you have something to say too.

0:38With RSS.com, starting your own podcast is free and easy. Upload an episode and we distribute it to Apple Podcasts, Spotify, Amazon Music, and more. Track your listeners, see where they're from, and start earning from ads just like this. If you've been thinking about starting a podcast, this is your sign. Start your new podcast for free today at RSS.com. Anthropic has just released Claude Sonnet 4.5. Apparently, it's best AI model for coding, and they're already known for their coding capabilities. So I believe this is a pretty important update. Jaden is much more in the world of coding app development, things like that.

1:17So he's going to break it down for us today. But before you do, Jaden, why don't you tell them about our school community? So every single week, Jamie and I post an exclusive video over on our AI Hustle school community. There's a link in the description if you want to join. It's$23 a month. We increased the price, but if you get it now, you can lock in that price. This week, we did a really interesting episode where I'm starting a new genre of podcast. I'm going to start doing business news more broadly. Anyways, in order to get the podcast growing before I will probably manually take it over, or maybe I'll hire someone as a podcast network to run this show for me, which by the way, if you want to run a podcast, if you got a great voice, send me a message.

1:56Maybe you could run this one. I've actually completely automated a process of a virtual assistant that is uploading 10 episodes a day generated by AI, but they sound exactly like my voice, talking exactly like me. And so far, the results have been insane. I'm at 150 ,000 downloads on the podcast. 90 % of people can't tell it's me. You may be doubtful that that is possible, but if you want to learn how I did it, you want to know the name of the podcast, the analytics, and exactly, but literally step-by-step everything I've done, which is interesting, unique stuff in there. go and check out the AI Hustle School community.

2:30It's our latest video. There's a link in the description. All right, let's get into what's going on with Anthropic now. So Sonnet 4.5, this is just a really good model for a lot of things. So it scores high in the benchmarks. It's, you know, their latest model. But what it really crushes at is coding. And this is what a lot of people are talking about. And a lot of people are really excited about. Myself, personally, because I have AI Box, which is like a no-code AI app builder. AI box.ai. And we use Claude code. So it's kind of their coding tool to do literally everything inside of our product.

3:03We do not, we do not do anything unless it is done by Claude code because we can 10 X our outputs. So whenever Claude comes out with a new model, it makes a huge difference for us. This Sonnet 4.5, we were really excited about because it's really good at what they said about it. And then I'll tell you what we're using it for. What they said about it is it crushes all of the benchmarks for coding. Claude says that it is capable of building, quote, production-ready applications rather than just prototypes. So this is a huge leap in reliability from previous models. Basically, in the past, you could build prototypes, and we would obviously work and make those prototypes actually function, but it is getting better and better at just one shot.

3:42You tell it exactly what you want, and it can build everything needed for that, the entire pipeline. And so we actually recently just integrated it into our into AI box we have this like we have this tool where you explain like a type of product you'd like to build like an application you'd like to build and we had an older version of Claude in the past that would basically link together a whole bunch of different AI models we gave access to like 40 different AI models which you get access to all those on AI box for 20 bucks a month we give it access to all of those and it would build the whole tool it would link it together it would make prompts for it would make a front end and basically would make you this whole application.

4:14And it was pretty good, but we are seeing, I would say it's probably three times as good since we've updated to this new 4.5 model. The apps are way better. The prompts need less tweaking. So in the past, it would create prompts for your tool. And sometimes I'd be like, oh, you know, like it basically does the thing, but like the prompt needs to change slightly to make the output a little bit better. It is getting better and better at just one shotting that whole process. So I am insanely excited about this. And I think a lot of people, even if you're not a developer, should be excited because the tools you'll be using will have this integrated.

4:46So your outputs on a lot of different AI tools are about to get a lot better as well. Yeah. And so, you know, I've messed around a little bit with lovable, which I believe actually runs on Claude, but yes, you know, and one thing I was super amazed with is just how I know nothing about coding, but I could say, Hey, this link is broken, fix it. And it would just, it would know exactly what I, what I want to do and then do it. So I'm hoping that, you know, with this update, and I don't know if you've heard, Jane, but if some of the backend problems that some of those like lovable has could be fixed as far as plugging into some of your data and backend stuff, or is it, I mean, I know people use Cloud for all that.

5:26So where do you see this like being used mostly? Is it like the backend coding, the design or the design upfront or kind of both? I think both of them. And what we've been really impressed by is we don't actually have a designer at AI box anymore because we have so much design precedent for our entire platform. I'm pretty design savvy, but we have a lot of precedent. So we'll tell it to make a new page and it will basically design the whole thing. And the design is usually pretty dang good. It's looking at all of our code base. So it knows what everything else we have looks like. So it can build it in the same way.

5:59So we've been really impressed by that. It's great at the design. One thing I'll say about lovable is I think it's actually powered by Google. If I like last checked, it said, you get it run by Google, which their Gemini coder, is pretty good. On the benchmarks, the SWE Bench is a software engineering benchmark. This is basically the most broadly known used talked about benchmark. Their new Sonnet 4.5 scores 82%. That's its highest tier. Its lowest is like 77%. I think that's depending on, you know, they give it different amounts of power. But in any case, it can get up to 82%. its previous Opus 4.1 could get up to 79%.

6:38So it's, you know, like 3 % higher. Sonnet 4 was at 80%. So it's higher than Sonnet 4. GPT-5 Codex is at 74%. So it's quite a bit higher than GPT-5 Codex. GPT-5 is at 72%, even lower. Gemini 2.5 Pro is at 67%, which is why I'm surprised that Lovable's using Gemini for a lot of their stuff. I think if they do upgrade to Sonnet or if they give you the option maybe to pay more to switch because I think it's maybe a little bit more expensive is why they're using Gemini. But if they let you pay or switch, I think you'll get better outputs. I use Lovable quite a lot for making front-end designs basically.

7:15And I will give you everyone listening one really useful tip, even if you're not a software engineer, if you want to build really incredible looking designs for your webpage, I highly recommend Lovable. It's an amazing website for doing that. The problem is they don't make like actual applications that can do something very well. And so I think Jamie, you've tried and had that experience. So that's what I'd recommend doing over on AI Box, my platform. But Lovable is amazing for making the front end pages. The one thing I will say that I wish I had done, I recently redesigned the whole homepage of AI Box using Lovable.

7:44You could go to AI Box.ai and see what that looks like. That was all designed on Lovable. The problem I had was I didn't really know what I was doing. I'm not a coder in any way. So I got it to look how I wanted. I gave it to my developer and he was like, okay, this is great. But our entire project, like the entire AI box company, I guess, used like 49 different packages. Or there's basically like packages you have to install to do different things inside of the app. And he's like, your front end page uses 75 packages. So like the one page was just crazy. And he's using all the stuff we didn't have before, which is really annoying.

8:21We typically don't like to do that. So he did it anyways and was like, okay, like because it's just one page on the front page, I can kind of silo it and make it not mess anything else up and we'll just make it its own thing. But in the future, what I would recommend doing is if you have a developer and you have a project, ask them what the dependencies are for the project. Just say, what are the dependencies of our project? What does our app use? What are the packages? And have them listed out for you. They can actually ask like Claude, Code, or any AI tool to go through the entire project to make a list of them.

8:50He gave me a big list of our dependencies and I now plug it into lovable and say, make this page, but only use these dependencies and these packages. And it will, it goes through and changes it. So I essentially had to rebuild. I just rebuilt the page and we're going to get it launched again, but it's now it's only using our packages and dependencies. So it's really fast to add it. And it's not going to mess anything up on the site. So anyways, if you are working with a developer and you're trying to make like landing pages or, or different pages or even pages for your app, make sure that you do that.

9:20And that will save you a lot of pain. Well, that went above my head, but thank you for sharing. I appreciate it. If anyone is making an app with a developer, you'll know what I'm talking about. Your developer will be grateful to you eternally. One interesting thing this article said is that this release, Sonnet 4.5, is launching less than two months after their latest model, which was Claude Opus 4.1. Basically saying the rapid production cycles are making it difficult for any company to hold a meaningful lead for very long. So all that to say is this whole coding world using AI is rapidly advancing.

9:53At this point, it looks like Claude is the best one on at least the benchmarks he gave. But new ones pop up all the time and new releases are released all the time as well. So it's a rapidly developing industry and it's important to stay up to date on it, especially if you are an entrepreneur. I got to give you one crazy thing that was said about this. So Hershey, I think he works at Anthropic. He said he's seen Claude Sonnet 5 code autonomously for up to 30 hours during early trials with some enterprise customers. And he said that while it was doing that, so they said they give it a project and it was coding for 30 hours straight.

10:33In that time, he watched it not only build an application, but stand up database services, purchase domain names, and perform SOC 2 audits to make sure the product was secure. So they're saying that it's like, they're like, look, everything this model can do is not even on the benchmarks, which yeah, a lot of this stuff isn't like SOC 2 audits or not on these software engineering benchmarks per se, but it could do a lot of stuff. It's really impressive. So I'm excited to see what people build with it. And I'm super excited that I now have it helping us code everything over at AI Box. So yeah, super excited.

11:05Hey, if you guys enjoyed the episode today, if you learned anything new about building apps with AI tools, about Claude and everything going on in the world of AI and software development, make sure to leave us a review on the show. It helps the show a ton get found by other amazing people like yourself. And also make sure to go check out the AI Hustle School community for a whole bunch of bonus exclusive videos and breakdowns so we don't share anywhere else. All right. Thanks so much for tuning in. We'll catch you guys in the next episode. Hey, everybody. Lady Luck here. And we're celebrating America's 250th birthday.

11:38Now, all summer long, I'm going to be celebrating by playing on SpinQuest.com, which is an American-owned social casino. It obviously features over a thousand slot games and live blackjack, live craps, live bubble craps. Head on over to SpinQuest.com. Get yourself a$30 coin pack for just$10. SpinQuest is a free-to-play social casino. Void where prohibited. Visit SpinQuest.com for more details. You're listening to a podcast right now. Driving, working out, walking the dog. If you're into podcasts, chances are you have something to say too. With RSS.com, starting your own is free and easy. Upload an episode and we distribute it to Apple Podcasts, Spotify, Amazon Music, and hundreds more.

12:23Track your listeners, see where they're from, and start earning from ads like this, even with just 10 listeners a month. If you've been thinking about starting a podcast, this is your sign. Start free at RSS.com.