make some retro game like F-Zero. Have some fun with this. Okay, but let's go back to our slides here. And we've shown that Claude is the best coding model right now. But here's a catch. Claude is almost 20 times as expensive as Gemini 2. 5 Flash, right? So you see here that Claude costs...many cases it's good enough. And furthermore, Gemini also has a larger context window than either Claude or ChatgPT. Okay, so I guess go for Claude if you want the best coding model and go for Gemini if you want the best price toerformance ratio for coding. All right. And I also want to show you some other things
Поиск по транскрипциям
Claude Code's remote control is finally here. Normally when we want to use Claude Code, we have to do so in a chat window like this on our local machine. Or we could obviously pull up the terminal, type in Claude, and then access it this way. But now we're able to type in this command claude remote
here, we're going to go over to this lefth hand side and click on the extensions button. And all you have to do is type in Cloud Code. It'll be this one right here that's verified from Anthropic. And then you'll just go ahead and install this. When you install it, it will prompt...your paid Cloud subscription. Now, this does have to be the Pro or Max plan because if you're on just the free, you don't have access to Claude Code. Now, once you've installed this, what it will do is give you this little orange button in the top right, which lets you open up Claude Code
local storage just like I asked. Here's where things get really interesting. Let's say I want to add a feature. I can just go back to Claude Code and ask for it. Let me try something fun. So, I have some prompts ready here and there's two lines and you're going...understood the entire structure, and now surgically adding the new features. An edit button on each card, a dark and a dark mode toggle. This is the power of Claude Code. It understands context. Here's the moment of truth. I'm just going to get it to open it again. All right, brought it over here. So, what were
into Claude Co-work to get it to do an enormous amount of work for me automatically. If you haven't used Claude Co-work yet, it's basically Claude code for your entire life. It takes the AI agent that Anthropic built, which is by far the most powerful AI agent on the entire market and allows...side project. I need to figure out new functionality for Creator Buddy, which is my main project. I need to write a newsletter about how to build apps and claude code. And I want to find AI companies to invest in. And I say the only urgent thing I'm trying to figure
Tailwind CSS for styling. You can add lotty animations. You can integrate custom fonts and assets. The motion skill is marked as critical for animations and sequencing. That means Claude code prioritizes these capabilities. It knows this is core functionality. It builds them accurately every single time. Now, let me show you a real example. Let's say I want...hours per week. Scale without hiring. Add smooth transitions. End with a call to action. Join 40,000 members. Use a modern color scheme with blues and whites. Claude Code builds that entire video. It creates the text animations. It times the transitions. It positions the elements. It handles the sequencing. And I can iterate on it instantly. Make
Claude's adaptive thinking feature. When you're building something simple like a contact form, you can use default effort. Claude works fast and gives you clean code. But when you're building something complex like a multi-user dashboard with RO based permissions, you switch to max effort. Claude thinks deeper. It considers edge cases, security vulnerabilities, performance optimization...data table below. The more specific you are, the better the output. Second, when using Claude, include context. Paste in your API docs, your database schema, your existing code base. Claude will use all of it to generate better code. Third, use effort levels wisely. For simple tasks, use default effort. For complex logic, use max effort. This saves
Deepseek V4 leaks are absolutely insane. This AI might destroy chat GPT and claude at coding. Million token context windows that can read entire code bases. Performance that beats everything else in testing. The coding wars just exploded. Okay, so Deep Seek V4 is coming in midFebruary and the leaks are absolutely nuts. This could flip the entire AI coding...world upside down. Right now, everyone's using chat GPT and Claude for coding help, but Deep See V4 might make them look slow. The leaked benchmarks show V4 beating both of them on complex coding tasks. And that's just internal testing. Wait until this thing goes live. Here's what makes this crazy. Deepseek is this Chinese
Look at all the content it's put into cell B2. It doesn't make any sense, but it's still talking with Jack GPT. Can Claude use claw code use this feature as well? Yeah, claw code has that Chrome as well, so it can use this too. I'm going to stop the task there. But yeah...only got permissions and shortcuts. We don't have a way that we can change the AI that we're actually using. All right, let's get Claude Code running. We'll test this bad boy out. Panov says Opus is crazy expensive for them to run, so it makes sense. It's using Haiku and explains bad results
tools. Number one, Cursor. The IDE that made developers switch overnight. $9 billion valuation for a reason. If you write code professionally, this should be your default. Number two, Claude Code terminal first coding agent. If you live in the command line, this is for you. Number three, Lovable/Bolt. Best no code AI builders. If you're not a developer
better output. So, you can see how fast Qua Alpha is versus Claude, right? This seems to be a lot quicker. Like this is pretty much finished the coding whereas Claude is taking his sweet time. Is that actually is Claude even coding anything? There says I've created an interactive water simulation but there's nothing here...open up the example from Quazar actually went off and did its magic. All right. So if we compare it side by side, Claude 3. 7 didn't even create anything whereas Quazer actually did. Let's test it out directly inside Claw 3. 7 Sonet inside the chat feature. Maybe we'll get better outputs here
Sonnet 4. 5. And I have heard that to be generally true. Many people are using Sonnet 4. 5 to plan out what changes should be made in Claude Code, for example. And then they switched to a lighter weight model like Claude 4. 5 Haiku instead to do the implementation of the code because it's faster and also...like more cost effective. Anyways, we're going to talk about the Claude Code
want to know what categories what conversations what thoughts lead to better engagement so you can double down on those categories so here we go again claude's now writing all the code right in the browser I didn't need to download any other software writes the code and in a second's going to start running it listen...more updates about monetization so this is really helping me get a direction on what type of content performs well this is an amazing new feature from Claude it is writing code and running the code right
like to be building my app, and you can be using any app you want to build this, whether you're a cursor fan, a codeex fan, a Claude Code fan. Say I'm building my app out inside of Claude Code. What I'm going to do is exactly what you see here is I'm going to have
there just seems to be a couple more bug fixes instead, so nothing really important. But yeah, if you are interested in mastering everything to do with Claude Code, then I cover everything to do with that in my Claude Code masterclass. There's a Black Friday sale going on right now. Link in the description down below. And also
more time you spend planning, the more likely Claude will succeed. And I always press shift tab to cycle to plan mode. And if Claude for some reason has to start coding, I always hit escape uh to cancel. So think of Claw like an overeager engineer, then you really have to ring all the time, right...next tip, which is to ask Claw to explore solutions with you. So the first thing I like to do in plan mode is to ask Claude to explain the code base and then I'll tell it you know I want to build this feature. Can you explore solutions starting with the simplest one? The whole idea
models working together as a team. Claude Code is now a sub agent inside Perplexi Computer, meaning the platform can write professional grade code automatically. GitHub CLI was also added, meaning the platform can fork repos, push code, and submit pull requests directly to GitHub. In a live demo, um, for example, computer forked the open call repo, coded...submit the PR all from a single prompt. And they've also added custom skills, voice mode, model council, and GPT for recently to perplexity. So Claude code alone is still the best for deep focus coding work. But Plexi Computer is the best for like end to-end workflows that span research, design, and deployment. But basically, you know
building a simple app, creating a tool for your business, automating a process. You either pay a developer thousands of dollars or you learn to code yourself, which takes months, or you use Claude Sonet 4. 5 and you get it done in minutes for free. Let me give you some real use cases. You run an agency. You need...some of you are worried. Can AI do everything? Will it replace developers? Short answer number. AI is a tool. It's not a replacement for human judgment. Claude can write code, but you still need to know what you want. You need to test it. Make sure it works for your use case. You need to understand your business
Confucious Code Agent with Claude 4. 5 Sonnet scored 52. 7% on a hard coding benchmark. Meanwhile, Claude 4. 5 Opus with a weaker system scored 52. 0%. Wait, what? A mid-tier model with a strong system beat a stronger model with a weak system. That's the whole point. The scaffold, the system, the structure around
then on top of these primitives. On top of this agent, we have a few things. We have a very nice light UI layer where you get to watch Claude code work. You see all the text fly by and we have this nice permission system that allows the agent to work and allows and kind of forces the human...very productive because you're just trying to figure out where things are in the codebase, what patterns kind of the team is using, things like that. And Claude code can kind of help supercharge that onboarding process. You can ask claude, hey, where is this feature implemented? Or since it's great at the terminal