prefer more visibility into what's actually happening behind the scenes. So for example, like if you just leave it on the default verbosity and you say to Claude Code because you're using it for something non coding related, because it's a good agent overall, you say something like research competitive pricing and then update the spreadsheet. Then...Claude Code will be as efficient as possible and then give you a final output. And you're kind of like, why did I get this output? Like, why is it like this? Essentially, like the model's process has been hidden after each tool call that it did and each information that it fetched online. And then you would
Поиск по транскрипциям
have any favorite little design polishes or things that, touches in Claude Code? - I definitely do. I really like the ASCII reticulating and thinking. I think those are such a great point of personality for Claude. And I also really, really like the different modes, how we've like outlined if you're in thinking mode, or planning mode...process of programming can be like a robotic thing. You know, you're dealing with lines of code and lots of characters, but when you're using Claude Code, it's almost like a different experience and it kind of elicits a different emotion than just like if I'm in an IDE and I'm just typing line after
least. I think DeepSeek Reasoner is kind of overcomplicating the problem. So maybe my prompt has to be adjusted for this particular model. DeepSeek seems to do okay in Claude Code. It probably will perform better in much smaller code bases. This is a monorepo with many different like segments to it. So it probably doesn't perform as well...because the context window is smaller. And because the Claude Code system prompt is probably not fine-tuned for DeepSeek itself. And when it comes to pricing of the model, it is pretty good value for money. And they also have a discount price for these like off-peak hours over here, which I find quite interesting. So ultimately
decided to make your own Figma. What did that process look like? I heavily used PI coding agent with my own set of extensions and mostly claude code opus 4. 6. The workflow is quite simple. I start one sometimes up to three work trees. And by the way, work trees are kind of like uh something that's becoming...keep on getting better and better. Now I've seen it with my own project at designcourse. com. I spent two months completely refactoring it within cursor and then claude code 4. 6 and I replaced a SAS of my own with literally within a couple days. Live chat. com I made my own live chat and guess what
content creation. It's exceptional at complex coding tasks. Game development, interactive applications, advanced programming challenges. When you need something that actually works, Claude delivers. 5 Pro are fast. But fast doesn't matter if the code is broken, if the functionality doesn't work, if you can't actually use what they create. This is the difference between...doing content marketing, this is your AI. The quality is consistently excellent. The engagement is superior. The readability is perfect. But that's not all. Claude also dominates complex coding tasks, game development, interactive applications, advanced programming. When you need code that actually works, Claude is your AI. Grock 4 is the analysis beast. Competitive research, strategic planning, comprehensive reporting
Google account. That's it. Run one command. Sign in with Google. Start coding. I'll put the installation instructions in the description. If you want to try Claude Code, you'll need to sign up for their beta. And you'll need a credit card because it's going to cost you. Look, this isn't really about...Goldie reads every single comment, so drop a comment below and let me know which tool you're going to try first. Are you Team Gemini CLI or Team Claude Code? And what are you going to build with whichever one you choose? I want to hear about it. And if you build something cool, I might feature
that I'm doing all of this is I'm doing this in a platform called Visual Studio Code. Then I'm also using um platform called Claude Code with my Visual Studio Code. Okay. So, if I open up my own instance for you, it looks something like this. I have a bunch of files on the lefth hand
going to hit enter on that. And now while we have agents working for us on Claude for Webb, while we have our agent building out functionality inside Claude Code, I now have like a general business and life consultant. I'll be talking to in the claw desktop app. And so you don't even need to just talk...avoiding downtime. As long as you use AI whenever you have downtime, you will constantly be productive and be getting better. And so I have my AI and Claude code building out the big tasks. I have my AI agents inside Claude for web doing the little tasks and the little planning. And then I have the clawed desktop
spending some time is in the coding tools. Not to write code, but because the coding tools have those skill in agent capabilities built in, so that's Claude code, that's GP open AI's Codex, that is Google Gemini code, uh, or Google's neutral anti-gravity, which might Rachel by the way, support skills...portions and stuff. And it is a rapidly involving product. They have released up three releases in the last week alone. It is built, it was actually built by Claude Code, which is kind of funny. Um, and essentially it allows very similar to what, uh, Rachel and I are talking about. It allows you to use Claude
have all kinds of different courses on the top AI tools. Claude is one. you haven't made a course about, but now with Claude Co-work, one of my favorites with Cloud Code, one of the best AI tools of all time, we're working on a really comprehensive course. You'll get access to that when it gets...cancel anytime. I'll put a link in the description to that trial along with everything else I mentioned in the video. And I'll have more videos on claude code, on cloud code work, on cloud skills coming up on YouTube as well. Thanks for watching. I'll see you next time
show you in a very meta way how I prepared to go on your podcast today and how I also transition from using claude standard to claude code to go through the entire like flow of this. Sound good? Perfect. Sounds great. — Cool. Let's jump into little baby cute Claude. Now, what I'm going to show you most...know from talking to a lot of CEOs over the last two weeks, most people still don't understand how to use claude properly, especially how to jump from cloud to cloud code when needed. So this
would say so. I don't think I even need to think about that. I would say like genuinely from what I've seen, Claude is really good for and powerful for coding. Plexi is great for automating your browser. Honestly, if I had to choose between one or two, I mean, it depends what your...case is. If you're using research or you're trying to automate your browser, I would use Plexi. If you are trying to just vibe code something out, you could use Claude. But the problem with vibe coding is like you probably you're probably still going to get better results with Gemini, right? As I've shown
would say so. I don't think I even need to think about that. I would say like genuinely from what I've seen, Claude is really good for and powerful for coding. Perplexi is great for automating your browser. Honestly, if I had to choose between one or two, I mean, it depends what your...case is. If you're using research or you're trying to automate your browser, I would use Plexi. If you are trying to just vibe code something out, you could use Claude. But the problem with vibe coding is like you probably you're probably still going to get better results with Gemini, right? As we shown you today
questions here. So let's see what we got. It says what is he working on? Ralph and claw code. Alson says coding with claw code regardless of claude opus consumes credit very fast. If I subscript to the plus van any better options. So it's the same thing, right? I made a Gemini gem for PowerShell running terminal...will go off and build separate tasks with multiple agents. That's one of the easiest ways to do it. You can set this up with claw code and automaker which is a free GitHub and then from there you've got a nice UI where you can host it and test it. Skully says Goldie Pop. That's what
going to open this up. This is an example of the page that Claude did with pretty much the same props right now. Claude has always been like my favorite tool for coding and that sort of thing. But look at the difference here. And don't even get me started on chat cheapy. I'll show you the examples...smart enough to figure out, okay, we need HTML, right? And so for me personally, you compare the coding quality of Gemini canvas versus Claude and Gemini's code is nicer. Like it the UI is 10 times nicer, especially on the canvas. If we compare it versus chat GPT, chat GPT doesn't even create the website, right
than chat GPT in terms of UI this looks slightly nicer a bit of color in the CTA button not great Gro is still coming back to us Claude has finished coding it now and that's exactly what we want again nicely designed Landing Page information about who we are clear workflow analysis and what we need...with a report like it blows everything out the water compare deep seek which is the code here versus Claude here exactly the same prompt totally different outputs why is that it's because the intelligence of the model right it's the power of the model imagine if you were coding how much of a different output
than chat GPT in terms of UI this looks slightly nicer a bit of color in the CTA button not great Gro is still coming back to us Claude has finished coding it now and that's exactly what we want again nicely designed Landing Page information about who we are clear workflow analysis and what we need...with a report like it blows everything out the water compared deep seek which is the code here versus Claude here exactly the same problem totally different outputs why is that it's because of the intelligence of the model right it's the power of the model imagine if you were coded how much of a different output
about Claude as well is that I don't have to tell it to use canvas. it just goes off and uses it. So, you can see here Claude is coding the page out. I would expect that to look really good cuz Opus is just goated for creating uh website designs. Like, it's super nice. If we have...saying right so chat GBT I have to tell it to code it's not intelligent enough to understand my user intent whereas for example claude understands that I'm trying to code and it codes out automatically without me having to tell it and the same on Google Gemini codes out the page without me having to tell
Today I'm going to show you the craziest AI coding battle of 2025. Claude SA 4. 5 just dropped and it can run for 30 hours straight. GLM 4. 6 came out swinging with 200,000 tokens of context. I tested both of them head-to-head on real coding tasks. One of them crushed it, the other...about what that means. You can give it a massive project. Go to sleep, wake up, and it's still working. That's insane. But that's not all. Claude Code got massive updates. It now has checkpoints, so if something breaks, it can go back. It has code execution built in so it can test its own code
also you want to focus on productivity instead. All right, so we're going to allow this to edit the settings so it can edit its own settings inside code here. And then let's say for example you've got a website like for example we actually built out a new version of the AI profit boardroom website which...automation community and we've deployed that using this chat right here. Right. So if we go to this link, we have this website deployed from claude code, right? And this was built using claude opus 4. 6 and fast mode inside claw code, right? So what we're going to do now is we're going to use agent