here. So over on the left hand side is going to be all the files you're going to create once you start giving prompts to claude code. Over on the right hand side is claude code. This is where you're going to give commands to the AI agent. Build me this, change that, debug this, launch that...claude code will do all the hard work for you. Again, you do not need any programming experience here. Claude Code's going to do all the hard technical work for you. Before we start building out our application, I'll show you other couple parts of the UI here. uh a lot of the configuration of claude code
Поиск по транскрипциям
weaknesses, then create a professional PRD document with charts and tables that include target user, core user loop, MVP scope, text stack, timeline, naming, domain suggestions, launch strategy, and claude code starter prompt. In just that one prompt, just those couple sentences right there will now send our 247 AI employee off to go validate the idea, build...opportunities, threats, targeted user. This is a 14page PRD. It created core features for the MVP and gave me the entire tech stack. That's amazing. Look at this Claude Code starter prompt. So, I can just take this prompt, give it to Claude Code, and it'll start building the app. If I want, I can just have
down below, put into Claude, it'll just build the skill out for you and this will build a new skill for us that we will plug into Claude code. So once our app is built out in a second, we can start generating the landing page and all the marketing content very easily. Again, if you would have asked...Claude Code to build out the marketing content, if you didn't have this skill, it would build out very generic AI content, but now it's going to build super powerful marketing content for us, which is going to be amazing. So, it's building out this skill. It even has examples of copy the AI should write
more tokens, is more work, wastes time, and fixing code with AI is not a lot of fun. And by the way, I cover in detail how to use Claude Code, how to write prompts, how to do context engineering, how to provide general rules, how to use tools Claude Code offers like agents or skills, to which...back in this video too, And I'll build a complete demo project in a brand new course I just launched about Claude Code. You find a link to the course with a great discount below the video in case you're interested. The third point here is that I use agents and skills, and what do I mean
lead developer relations at anthropic and today I'm going to be showing you a coding task with computer use so we're going to be showing Claude doing a website coding task by actually controlling my laptop but before we start coding we need an actual website for Claude to make changes to so let's ask Claude...want to make a few changes to the website locally on my own computer let's ask Claude to download the file and then open it up in vs code Claude clicks the save to file button opens up vs code and then finds the file within my downlo loads folder and opens it up perfect now that the file
seems good at SwiftUI. But I think for the other use cases, when it comes to Next. js, especially like I did with MindDeck, I will continue to use Claude Code. It's quite interesting that each model kind of has its own strengths and weaknesses. But I think overall, I will continue to be using Claude Code because just...slightly different things. So some people will have multiple coding agents. They might use GPT-5 when it comes to some kind of design work and they might use Claude Code for bigger tasks or they might want to like optimize their costs in some way. I don't think there will be an overall one winner, at least
context from claude. md. This is our business context that I created. So this is going to be what I'm going to copy and paste into claude code. Again make sure you change the folder here to the new folder here. I'll press autoac accept edits and we'll go ahead and send this out this time...change next, which comes usually in the iteration in the follow-up prompt. And that's it. That's typically the structure of any prompt you give to Claude Code. Okay, here is the app that has created for us here. So, it's pulled in the latest AI news here from different sources that it picked automatically
make. Now my class also has a MCP server, which is a really good use case for forked subagents. So if you add like the MCP server to Claude Code, I can basically tell Claude Code, with a forked subagent, can you use the Agentic Coding School MCP and recommend me any videos to watch that you think would...most helpful based on the session that we've just done together? Pressing enter, Claude Code will use all the nuance and detail of the current conversation with a forked subagent. It will do all the noisy tool calling in the fork to gather all that information and then just return that back to the main session
down below. Leave a like. All I do is make amazing videos on how to build with AI. So, what's going to start happening now is Claude Code is going to go and start building this plan for us. So, it's going to start thinking and designing what it wants to do, and then it's going...plan together, then we can have it go and start building us out the application for us completely autonomously. But this is one of the really big strengths of Claude Code is the planning it does and just the highlevel thinking is unmatched by any other AI tool I've seen on the planet. Especially if you're using Claude
that we vibecoded. Towards the end of 2025, we knew that our site needed a redesign. Me and my co-orker Molly decided to prototype a new homepage using claude code. Neither of us are developers, but we figured it would be a good way to see what the new site could look like. 8 hours of prompting later...started migrating page by page, collection by collection, item by item. Since our Web Flow site was already live, building the static pages was as simple as sending Claude code the URL of the live page and saying, "Rebuild this in the style of our new homepage. " 100 plus pages later, everything worked and it all went by with
building stuff. That's the best compliment I can get. It's the best in the world. Um, all right. Let's go back to Claude Code and let's get this live, huh? Mostly on Forest, my man. You're only a couple minutes late. I like it. You're forgiven. Honest question here, Alex. As vibe coding...experience, that's where the academy is. That's where the community is for. I've been doing a project as an experiment where I use claude code in the terminal and have a finder window of the project open. I don't need to look at files. This gives me all the info I need. I'm actually considering
Today we're going to be testing out Claude code with O Lama. So essentially what you can do now is you can plug in Olama directly into Claude code to run Claude code and other tools with opensource models. So we're going to test this out and see how it performs. You can also use cloud models...local models on your machine and then run these directly through claude code. This is how you get started. These are instructions. We're going to just test this out and see how it performs. All right. Now, the first thing you want
comparing against Claude for Sonnet using Cloud Code and also Gny 2. 5 Pro using GM CLI. So in a previous video, I compared Claude for Sonnet using Claude Code to Jimmy CLI over here. In this video, Claude Code did better than Jimmy CLI. So I'm going to give the exact same instructions...doing this is basically that Claude code can now actually support Kimmy. So Kimi
chat interface. But of course I think for many scientists, maybe some that do, maybe some that don't realize this, the power of agentic coding tools like Claude Code or other places where that longer context and all of that power becomes really interesting for data analysis, for integration, for kind of like reasoning over different types of knowledge...start to build. - Yeah, it really is, and I know this is one of the things that you and I have been the most excited about, that Claude Code is amazingly useful as it is today in biology. And most people don't realize that, right? It's called Claude Code, it's not called Claude Biology, right
next trick you need to know about is saving your spot with checkpoints. So, the number one question I get about clawed code is, "How do I rewind if Claude code messes up? " People are used to cursor. When you're in cursor and you're using the cursor agent, you can easily rewind by clicking on checkpoints...commits a day. If Claude messes up, I'll go back to my last commit. So, I'll rewind to that last commit and start over again with Claude code. This is how you make up for the lack of checkpoints like Cursor Agent has is by using GitHub as your checkpoint system. If you're a new programmer
with Claude Opus? That means Claude can process images better and faster than any other model and they can do multiple images at once. So how does that impact claude code? This means you should be using as many images as possible with every single prompt you use. So there's two main reasons you'd want to use images...prompt with Claude code. One is for inspiration. So say I want to take inspiration from another app I'm building. my fantasy football app. All I would need to do is screenshot this and say put this into our prompt. So paste it right into the terminal. You can paste images right into the claude code terminal
using slash cost. This helps you check your current token usage. You can also configure a status line to display continuously. In order to configure a status line in Claude code, you just go slash status line. You'll notice that you can't currently do this inside of the GUI version. So what you have to do instead...tokens that you give it to like maximum think from, I don't know, 8,000 to like 32,000 or more. Now agent teams is another feature of Claude code, which I'm looking forward to covering with you guys, but currently it costs a lot, about seven times more tokens than standard sessions, especially when teammates
Claude Code inside a cursor. I'm going to give it the same exact prompt. I'm going to hit enter. Let's see how Claude Code does here. I've built many games with Claude Code before. It handles 3JS really, really well. So, I'm expecting good things here. I was impressed to a point with the GPT5...from kind of a stylistic vibe perspective. From a total execution perspective, it could use some work. One thing I love about Claude codes, it always adds a little bit extra. more than what you ask for, which I actually like. It seems with GPT, it just added kind of the bare minimum. It didn't really have taste
your homework. But to help you understand, it can break down complex topics into simple pieces. That's how you actually learn. Let me tell you about Claude Code specifically. Anthropic announced updates to Claude code, including an upgraded plan mode that creates more precise plans for how to solve a problem or add a new feature and sticks...them more directly. Better planning means less wasted time, less going in circles, more getting stuff done. And Claude Code has been added to Anthropic's desktop apps, including the Mac version. It allows software engineers to code, research, and update work with multiple local and remote sessions running at the same time. You can work on multiple projects simultaneously
price, you're now going to get three times the usage by using Haiku 4. 5. And this, in my opinion, is really solving three things, which is one, Claude Code's too slow. It is a little bit slow. It is slower than the competition at the moment. Haiku is significantly faster than Sonnet. So you're going...good news is if you're using Haiku, you're going to get way way more usage. So you're going to get a lot more usage out of Claude Code. So it solves the Claude code is too expensive problem. And then three, Anthropic really needed an agentic model. So, everyone's building different AI agent frameworks. It seems