become easy. - One interesting thing I've seen here recently with a feature that we released in Claude AI on the web was the ability for Claude to create actual files through writing code. So it was like writing a Python script and then the Python script got ran and all of a sudden you have like a Excel sheet...SVGs, but then I wanted it to make a much more detailed diagram that would need a lot of repetition and so Claude was actually able to do this by writing some code to generate the SVG, which ran much, much faster than Claude itself needing to write you know, it was a very, very repetitive image file with lots
Поиск по транскрипциям
whereas Gemini 2. 5 Pro you can see it automatically goes straight into thinking mode before it starts coding anything out. And you can see now it's coding out. Claude is doing the same thing. And we'll see which one comes up with the best outputs. So you can see here, let's have a look. This...much detail as we want inside this project and then deploy it as well in a subdomain. But it's miles above the second best coding tool in the world. And I think until Claude releases like version 4, it's just not going to be able to compete. It's nowhere near the same level. Now, the other cool
harder to solve complex problems like debugging code or solving a tough math problem Claude 37 Sonic can now do both giving you quick answers when that's all you need or taking its time to work through harder problems when required and here's the crazy part you can actually see its thinking process yes you can watch...into all of its capabilities and showing you how to use them to supercharge your business let's get into it first let's talk about the coding abilities this is where Claude 3. 7 Sonic really shines according to anthropics announcement is now the best in class for real world coding tasks companies like cursor cognition versal replit
smash that like button, subscribe if you haven't already, and I'll see you in the next one. New Claude Sonet 4. 5 destroys chat GPT in coding tests. Claude just dropped a bomb that nobody saw coming. This AI literally builds full games from scratch in minutes. I'm talking real stealth games with AI enemies and sound...path. I'm not talking about small improvements. This is a complete gamecher. Let me hit you with the facts first. Claude Sonet 4. 5 just scored 82% on the swench verified test. That's the hardest coding benchmark out there. It beat Opus 4. 1. It beat Sonet 4. It even destroyed GPT5 codecs which only
stream here, this thumbnail was actually generated on the live stream using Nana Banana. So, it's too good. It's too good, my friends. Chat GBT is still coding out right here. Jayen, oh, good to see you, Jaden. It's been a while. Are you doing any more test sites in private? The good old days where people...anger's uh question you can create it on chat GBD does it look as good I wouldn't say so I do think as well claude is pretty good for coding if you want to use uh claude instead but I think out of all these just use Google stitch and then after that
because of two major releases that Claude has released over the last few weeks. The first one is Claw Code Remote Control, where you're able to use Claude Code from your mobile app. And now they've released Claw Co-work Dispatch, where you're able to do the same thing with Claw Co-work. Cloud code...developers and people who are familiar with and feel comfortable working with code and commands in the CLI. Claw co-work is for people who like a typical graphical user interface where there are buttons to click on labels and it's just plain text easy to use tell claw what to do. And in this video I'm going
input at all. This is a huge shift. And if you're not paying attention to this stuff, you're going to fall behind fast. So, let me recap. Claude Code now has scheduled tasks. You set a prompt. You set a schedule, Claude runs it automatically. It can fix bugs, review code, update packages, write reports, monitor systems...want the full process, SOPs, and over 100 AI use cases like this one, join the AI profit boardroom. Learn how to automate your workflows with tools like Cloud Code and scale your business with AI. The link is in the comments and description. And if you want free access to all the video notes, prompts, and a community
right, so a couple of days ago, Claude Sonnet 4. 5 came out and it claimed to be the best coding model in the world. A bunch of people have tried it and some people thought it was really great and other a bunch...crap and there had been a bunch of mixed reviews in the middle. I'll be sharing my own thoughts about Claude Sonnet 4. 5 having used it in Claude Code for about 20, 30 hours or so and then also sharing the experiences of other people online as well. So the first thing
integrate to antiggravity sometimes poor internet. Yeah, 100%. So here's what you do. So it's like you're using because anti-gravity is just like virtual studio code, right? And so what you can do is you can run these local extensions inside anti-gravity, right? So for example, client which you can see right here allows...drag it up and down like that. And then from here you would just type in whichever model you want to use. So if you want to use claude code for example, you would just type in claude inside the terminal here and then you can start using claude to enhance your project as well. Right
integrate to antiggravity sometimes poor internet. Yeah, 100%. So here's what you do. So it's like you're using because anti-gravity is just like virtual studio code, right? And so what you can do is you can run these local extensions inside anti-gravity, right? So for example, client which you can see right here allows...drag it up and down like that. And then from here you would just type in whichever model you want to use. So if you want to use claude code for example, you would just type in claude inside the terminal here and then you can start using claude to enhance your project
because this thing is a massive upgrade. Anthropic, the company behind Claude, just released this, and honestly, it's one of the biggest jumps I've seen in an AI model update. Let me break down what's new. First, the intelligence. Sonic 4. 6 is way smarter than the older versions. It's better at reasoning, better at planning...model is a beast. It writes cleaner code. It catches more errors. It understands what you're trying to build and fills in the gaps. And inside tools like Claude Code, it can work like a real coding partner. You give it a task, it plans the steps, writes the code, and helps you debug. That's not a gimmick
more you start to realize there are a couple of gaps with it. And the biggest gap for me is reviewing all of the output from Claude code. Like we're looking at right here. Claude is ripping through a bunch of changes for me. And that's a beautiful thing. But when it comes time for me to review
want to check out my brand new master Claude Code class, there will be a coupon code down below with a link. And if you do want to check out the programs that I personally make in Vibe Code, like HyperWhisper, there will be links and coupon codes for those down below. Because I don't accept any sponsors
gets all the console logs from that and can deliver a report to Claude code on what doesn't work so that Claude can fix it. And as I added that, it's already telling me, okay, since you added a QA breaker agent and added Playright, these are great tools for testing your no machine...just going to close out and open up new version of clot code. And now the MCP should be active. So if I say /mcp, ah, still not active. So I'll run the command once more. And I think I need to restart Claude to see that. There it is. Play right. Perfect. So now I can say / agents
this is the first step just creating a new repository. I have this folder. And because I went through the setup checklist in advance and I already have cloud code that is working on my machine. I will just rightclick this go to services and say new terminal at folder here on the Mac. This opens up a brand...going to be doing at this very point. Okay. And the only thing you need to be doing is saying claude. Hit enter. Beautiful. Now it initializes claude code and it asks me, hey, can I read files from this folder? Because I did this little thing with the services and open terminal at folder. The terminal is already pointing
schedule section here, you can see that we can set up schedule tasks, but now you can run it inside claude as well. So if you go to code, then you go to scheduled, you can schedule whatever you want, right? Um, and it's really cool. So you could say, for example, okay, you know, uh, I create content...improve my business or each day do one thing and automate it whilst I'm sleeping that would help me grow my business. " And you can just get Claude code to do that in the background. So it saves a lot of time. It's running automatically. And also
least free way to support this channel is sign up for the Vibe Coding Academy link down below. has a full Claude code course in it. Full uh Xgrowth course, weekly calls. We're gonna do a weekly call in eight minutes. I'm joining. I'm answering questions. We're chatting. We do hackathons every month. Uh hu huge
make a modification to a prompt and then press Enter again. Unfortunately, I have no idea what this "unshipped claude config command" is. But for the Claude Code
changing all his plans, and how you can get Kimmy K2 set up so you never have to pay for AI coding again. This might be the end of Claude Code. Let's lock in and get into it. So, this is Kimmy
models would sometimes generalize from one type of misbehavior to other misaligned things. We saw there was a prior research project that external researchers did on training on vulnerable code, but then what we saw from those projects was not that strong of a generalization. Like the models would just do misaligned things occasionally, but us it was really striking...only thing that we saw, right? We saw the model, you know... We did the most realistic thing we could, you know, we put it into Claude code, we had it work on the code for this research project and it decided, you know, we should sabotage this research project. The model decided it would actually attempt to make