Hello and welcome to the ultimate Claude tutorial for 2026. In today's video, I'm going to be teaching you everything that you need to know about Claude so that even if you are a total beginner, by the end of this video, you will be an expert using Claude. We're going to start with the basics like prompting Claude and configuring your account and then we'll work our way into some of the more advanced features like using co-work and using Claude code.

If this is your first time watching this video, I recommend watching it all the way through at least once and then coming back and replaying certain chapters if you need help with something more specific.

First, let's talk about pricing. So, if you head over to claw.ai and then you hover pricing in their menu and hit overview, that's going to bring you to their pricing page. And if you scroll down here, you'll see that they do have a free plan that you can get started with, but it is pretty restrictive. I'm not going to lie. It's enough to test things out, but you're probably not going to get any serious work done with the free plan. What I would recommend is starting out on the pro plan and then eventually upgrading to Max if you notice that you're actually getting a benefit from this. You can think of Claude as really an extra employee. And when you think about it that way, it's actually quite cheap.

For today's tutorial, I'm going to be starting out on the free plan so that you can see the limitations of it and you can decide from there if and when you want to upgrade. So, I'm just going to hit try Claude. And you can either enter your email or continue with Google. I'm just going to continue with Google. And then I'll choose my account and hit continue. Then to create an account, I have to accept their terms and hit create account. Next, I'll enter my phone number and I'll send a verification code to get signed in. So, they've texted me that code. Now, I'm just going to enter it and I'll hit verify code.

Then, it's going to ask me, how are you planning to use Claude? For me, I'm just going to select for personal use, but you can always say with my team or for personal use. And then you can come back later and change this setting. And then they're going to show me the plans again. For now, let's just start out with use Claude for free. And then it's going to mention Claude desktop, which I'm going to be signing up for later in this tutorial. But for now, we're just going to stay in the web version. So, I'm just going to hit skip.

And now, here's a key consideration for your privacy with Claude. You can turn this off if you don't want them to use your data and your chats that you have with Claude to train their models. So, I'm just going to turn this off and I'll show you how you can turn it back on if you want to later. But for now, I'm just going to hit continue. And it's sort of going to onboard me now. It's just going to ask me a few questions and I'm just going to type in my name and I'll go ahead and send that off. And then it's going to ask what I would like to use Claude for. What kind of topics am I interested in in my work and in my life? So, I'm just going to say life stuff, business and strategy, and coding and developing. And then I'll say let's go. And because I said coding and developing, it's going to recommend that I use Claude Code as well. And I'm going to show you how to use Claude Code later, but for now, I'm just going to say maybe later so that I can continue showing you guys the rest of the web app. And now, it might give you a few ideas for starting your first chat with Claude. But I'm just going to say I have my own topic. and then it's going to let me into my Claude account.

In the center of the page, you'll notice that you can simply type into the chat bar and you can ask Claude a question or give it a task. On the left hand side, you can see that this is where your chats are stored, your artifacts, your projects, pretty much everything that you're going to be using as a history in Claude. And then you can also customize and search. I'm going to be talking more about this in a moment, but you can also hide this sidebar if you want a more focused view. And I'm going to open this up really quick. And we're going to start out just by going down to my name down here and clicking on it. We're going to open up the settings in Claude. And I'm just going to go over the settings so that you understand where everything is. And you can set up Claude to best serve you.

So, starting out here on the general tab, we can edit our name. We can also reload the profile picture to get a nice profile picture that Claude provides for us. So, I'm just going to go with this one. And then it asks what best describes your work. And it gives some different options, some different job titles or roles. And you can just select your role. For me, I'm just going to select marketing. And then you can start entering some personal preferences here. So, this is just going to tailor your experience with Claude. You can write about your job, anything about your life, anything that will be relevant in your conversations with Claude. For me, I'm just going to be using voice to text. It's a bit quicker for this tutorial. So, I can click into here and then I can just use voice to text. I'll just say I run a YouTube channel called Productive Dude. If you're curious how I pulled this off and how I'm using the voicetoext feature, this is actually a separate app called Whisper Flow. Now, I'm just going to save changes to make sure that this stuff gets saved.

And next, let's talk about notifications. So, you can have notifications for response completions. This means when Claude is finished answering your question or your query that you send it in a chat, this will notify you. Remember, this is useful for longunning tasks. Like if you ask it to go do a bunch of web research and you don't want to be sitting there waiting for it, you can get a notification when it's complete if you turn on response completions notification right here. You will have to allow that in your browser. And now it will send me notifications.

Now the next notification is dispatch messages. And dispatch is a feature that I'll talk more about in the future, but it's basically giving Claude access to your computer while you're on the go, while you're away. And dispatch is how you talk to it. So, you can talk to it directly through your phone, and it can use your computer to do different things with your file systems, use your apps, and really, it just makes Claude more productive because it has a computer harness. It's not just chat and web search. It's able to actually utilize your computer, your file system, and your apps. I'm going to leave this off for now and we might talk about it a little bit more when we get into dispatch.

Now, I'm going to scroll down a little ways here and there are a few styling considerations that we have. For the color mode, you can go with light mode or dark mode. For this tutorial, I'll stay on light mode just so that it's easy for you guys to see. You can also change the setting for background animation. I'll go ahead and enable that for this tutorial. And if you scroll down a bit further, you can change your chat font. So, you can stick with their default. You can switch to SANS if you prefer that or you can use system or even dyslexic friendly if you're dyslexic. This can be helpful for reading the chat responses. I'm just going to stick with default. And then the final setting within general is the voice settings. So if you're actually talking to Claude, they have a few different voices that you can listen to and you can decide which one that you like. I'll talk more about voice mode later in this tutorial.

The next settings are your account settings. So you can click on account right here. You can see your active sessions with Claude down here. So, if there's anything suspicious, you can log that session out. This is also where you can log out of all of your devices if you want to. You can delete your account and you can access your organization ID.

Next, let's go to our privacy settings. And from inside of privacy, you can turn off your location if you want to so that Claude can't see your location. But this can actually be helpful if you want to keep it on so that if you ask it something about a local company or a local restaurant, it's able to actually answer that question with relevant information even if it doesn't have your location in the text. And then next we have help improve Claude. This is the one that we turned off when we were signing up for Claude, but you can turn it back on here if you want them to be able to train on your conversations. I recommend always having this off. There's no real good reason to have this on. Then here you can also export your data. You can manage any shared chats that you have if you share chats and you can also manage your memory preferences here. These are the things that Claude is going to remember about you.

Next tab that we have is billing and this is where you can upgrade your plan in the future if you want to. And you can also see invoice history and things like that. Right now I'm on the free plan so the billing page is pretty boring.

Next we have capabilities. And this is where you can really get Claude customized for you. If you want Claude to remember things about you as you chat with it and be able to generate those memories from any conversations that you have, you can turn this setting on right here and it just instantly enhances the power of Claude. You can also import memory from other AI providers. And to do this, I'm just going to use chat GPT from OpenAI as an example. I can just hit start import. And what they're going to do is they're going to give me this prompt right here that I can export my memories with. So, if I just copy this prompt, I can head over to something like chat GPT and if I've been using it for a while, it'll actually have a lot of memories stored about me already. And I can just paste that prompt that Anthropic gave me. And then I can go ahead and send this off so that we can pass that information over to Anthropic. So, I'll go ahead and send this off. And from here, it begins generating any memories that chat GPT might already have about me so that when this is done, I can just simply copy this and I can hand it over to Claude. For the sake of this tutorial, I'm not going to be copying over all of my memories from my actual chat GPT account. But if you want to, you can actually copy the memories that it gives you, and you can just paste them right here, and that will actually transfer all of your memories over. Once you hit add to memory, for now, I'm just going to close this import memory tab.

The next setting is tool access. We'll be talking more about adding tools to Claude later in this tutorial. But what this setting does is it allows you to load the tools when needed. So Claude can kind of see a surface level idea of what tools you have and then it can decide, okay, I want to use this tool. Or you can start with all of the tools loaded in so that Claude can see exactly everything that those tools can do. The only problem with this is it's going to eat into your conversation limit. So, you don't want to turn on tools already loaded if you're trying to be efficient with your token usage. I recommend keeping it on load tools when needed because Claude does a good job at figuring out what tools it needs rather than loading in all of the tools with this option here. This is going to allow you to have longer chats with Claude and it's not going to constantly fog the conversation with every tool that you have.

Next, we have visuals and this is where you can decide if you want artifacts. Artifacts are basically just mini apps or code that Claude can generate and render within Claude. So you can build these mini apps and have Claude run them directly on this platform. You don't need to have your computer or anything like that. You can just see and use them right inside of Claude. Now, if you want those artifacts to have AI inside of them, then you can actually have Claude build those artifacts with itself embedded in them. And to do that, you have to turn on AI powered artifacts. For now, I'm going to leave this off. and we may implement AI inside of the tools that we're building with Claude later down the line. Another feature that Claude recently added is inline visualizations. So, an artifact you can think of as a bit separate from the chat. It builds the app and it puts it almost in a side panel that you can then come back to and it saves it there. Whereas inline visualizations actually generate code inside of your chat so that you can visualize it right in line with the chat. For me, I actually prefer just sticking with artifacts and turning this off, but you can keep it on if you want, and I'll keep it on for the sake of this tutorial, just so you can see how it works later in the tutorial.

Then we move on to code execution and file creation. If you don't want Claude to be able to execute code, you can actually turn this off. And then you can turn artifacts and inline visualization off as well if you don't want it generating code. But one of the most exciting things about Claude is that it can do this. So, I'm going to leave this on for now. And then you can also decide if you want to allow network egress. This is where Claude's going to install code packages that make its code more powerful. If you want to turn this on so that you can enhance your artifacts and your inline visualizations, then you can do that, but make sure to check out the security risks before doing this. For now, I'm going to leave this off.

The final setting that we have here is skills. And I'm going to be going in depth on skills later in this training.

The next settings page is connectors. And this is where you're going to connect to different tools that you have. So you can browse common connectors. You can connect directly to Google Drive or GitHub right here. Or you can add a custom connector. And this is what they call an MCP. MCP stands for model context protocol. And all this really means is it's a way to connect outside apps to Claude so that it can utilize, read, and write to those apps. If you hit browse connectors, you can see all of the popular ones that they have that are just a one-click easy install. There's probably some popular tools in here that you're already using. And we'll talk more about this later. For now, I'm just going to close out of this.

And finally, we have Claude Code at the bottom here. Now, if you're not on a Max or a Pro plan, then you aren't able to use Claude Code. I'll be upgrading later in this guide to show you the power of Claude Code.

Next, I'll go down to the lower lefthand corner, and if I click on my name right here, I can then decide what language I want to use. I'm going to leave this on English, but you can change this if you want to. If you need to get help, you can hit get help. And this is going to open up Claude so that you can actually chat with it through a messaging portal if you want to ask a question about your account. I've had my account go down before and it's always good to be able to talk to Claude directly and figure out what's going wrong. Claude is a super popular app and it's in high demand, which means that sometimes Claude can go down. You can go to status.claude.com claude.com to see its current uptime and to see its uptime by platform. So claude.ai is what we're using right now and it's actually in a partial outage right now. I haven't tested it out so I don't know if we're running into issues, but this is just something to keep an eye on if you notice that Claude isn't working for you.

Now, you can also go in and upgrade your plan from here. You can get apps and extensions. You can even gift Claude to someone if you click gift claude, which is actually pretty cool. You can give someone access to x amount of months on a max 5x or a max 20x or even just the regular pro. If you're struggling to find a birthday or Christmas present for someone, this actually might be a great gift. I'm going to close out of this for now. And then finally, you can also learn more down here. They have tutorials, courses, you can learn more about Anthropic, you can read their policies and such. And you can also see their keyboard shortcuts here. If you're somebody who likes to know all the keyboard shortcuts and just know how to navigate everything quicker. I'm going to close this for now. And the last thing to touch on from this bottom sidebar here is the log out button. So you can just simply log out here.

Next, let's talk about the chat box. So here underneath on a new account, it's going to give you a get setup with Claude and it's going to give you some items that you can check off. I already showed you how to bring your history over from another AI. So, I'm going to close this one. And we're going to talk about setting up projects with files in context and also teaching Claude your workflows with skills. But for now, I'm just going to dismiss the rest of this checklist.

Next, you can click on the model selector right here if you want to change your model. Some of these models are better for quicker everyday tasks, while others are better for heavy reasoning or development or any kind of big gnarly task that you need to tackle. Opus 4.6 6 is my daily driver and I use it for some pretty heavyduty tasks, but this one isn't available to the free plan. But the good news is is Sonnet 4.6 is almost just as good and this is the default on the free plan. You can also go down to Haiku if you want faster answers and you want to use up less of your tokens. You can also turn on extended thinking if you want the model to think for more loops and you want it to think deeper about your problem. This can be good if you have a big problem to tackle. And then you can also hover more models and you can see some of the other models, some of the older models that they have available. Trust me when I say that this is a super fast-paced industry. So by the month, these models in here are constantly changing and they're always developing something new. So that's why I like to be upgraded typically. But for right now, we're just going to stay on set 4.6. But just keep an eye on these different models that they have available. Sometimes a model won't be able to do what you want it to do. You'll come back in a few months and then you'll be very surprised with the results of the latest model.

Next, we can create connections and attachments for our chat by clicking this plus button right here. And inside of this plus button, you can add files or photos from your computer. You can take a screenshot. You can add a chat to a project. And we'll be talking about projects in a bit here. You can also connect to different skills. And there's already a skill in here called skill creator that Anthropic provides. You can also add connectors, which we already talked about. These are the different apps that you can connect to Claude for reading and writing to using Claude. And then we also have web search. And by default, this is enabled. This is going to allow Claude to use web search to get relevant information from, let's say, today. If I asked it for today's news, the model wouldn't know that by default. What it would do is it would use web search if it had that available to it. So this is super important to keep on because it's going to make sure that all the information that you're getting from Claude is relevant.

And then we can also change the style of writing. I recommend leaving this on normal, but you can experiment with learning, concise, explanatory, formal, or you can even create and edit different styles that you have custom for Claude. So, let's say that you have your own writing style. Well, you can create that writing style here so that you can have Claude write exactly like you. When I hit create and edit styles, I'm just able to hit create custom style. And then here, I can add writing examples. So, if I just click this button, I'm able to upload from my device or paste text content. And if I don't want to upload that stuff, I can even just click describe style and I can explain the style here and go through the form. I'm going to close this for now and we're just going to stick to the normal style for this guide.

Now, let's say that I wanted to talk to Claude through my microphone and I wanted to hear it talking to me back. What I could do is I could click on the voice mode button right here. And this is actually really useful if you're in the car on your phone, you're going on a walk or something like that and you just want to talk back and forth without having to type into a keyboard and you can actually listen and talk back with Claude. So I'll just show you quickly how that works. I'll click use voice mode and I have to allow this. Hey there Claude. Can you hear me? >> Hey Carter. Yeah, I can hear you just fine. How's it going? And as you can see, it goes through and it continues with the conversation. So, as I'm talking here, it's basically pulling up my text immediately. And then when I stop talking, I don't have to do anything. It's just able to continue talking back to me like so. That's brilliant. So, you've got a seamless back and forth going here. No need to hit any buttons or wait around. Feels pretty natural, doesn't it? what would you like to chat about? >> For now, I'm just going to hit stop and then I will go back to new chat in the upper leftand corner. And that brings me back to this page. But now you can see that we have our first chat over here in the lefthand sidebar. If I want to, I can click the three dots here and I can star it if I want it to show up at the top of my chats. So if I'm having a really good conversation and I don't want to lose it, I can always star that and it will show up at the top of my conversations. I can also go in here and rename this so that I can make sure that I can find it. So, I'll just rename it to voice mode. And now I have that really nice and organized. I can also click into here and add this to a project. Again, we're going to talk about projects in a bit here, but for now, I'm just going to delete this because this was just a test conversation. So, I'll hit delete.

And if you want to use the clawed models without it having prior knowledge of your memories or without it having any sort of understanding of who you are, then you can click this little ghost icon in the upper right hand corner. And what that will do is it will open Claude in incognito mode. And as it says here, incognito chats aren't saved, added to memory, or used to train models. Now, this function is the same. I can just type anything that I want here, and I can send it off. But in this version, it's actually just going to contact the model directly. It's not going to give it any of my other information. I'm just going to hit X to close out of incognito mode.

And now, let's go over a brief prompting guide. No matter what large language model you're using, if you don't have the proper prompt structure, you're not going to get the best results. So, I'm going to briefly share with you the bulletproof prompting sequence that we teach to all of our students. Because if you have generic prompts, you're going to get generic answers. you're going to get the same answers that everyone else does. So, I'm going to share this prompting sequence with you and then I'll show you the results of a generic prompt versus the results of a prompt that actually does well. So, let's jump into the bulletproof prompting sequence. And this is an acronym that we like to call GCAO or goal, context, action, and then output format.

Here's an example of a good prompt. So, it says, "My goal is to create YouTube videos that will perform well for my channel. The analytics from my past videos are attached as context. I want you to analyze my past video stats from my channel. Then research my niche and share three videos that would perform well for my channel based on current trends and what has worked for me in the past. Your final output should be a bullet list of those video ideas with title, length, target, and a brief summary of what that video should cover." And this is a good prompt because it has my goal. It has the context. It has the action that I want the model to perform. And then finally, it has the output format. How I want to see that finalized data. Do I want to see it in a bullet list? Do I want to see it in a table? Do I want an interactive diagram? That's what I'm defining there.

So, first let's test out a generic prompt in Claude. I'm just going to say, I want you to give me ideas to grow my YouTube channel. And then I can send that off. And this is a short generic prompt. and it's going to give me a generic answer. The only context that it has is that my YouTube channel is productive dude. So, it might actually be able to figure that out. And as you can see here, it's actually asking me some questions. So, I can just select one of these options here. I'll just say productivity tips and tools. What's your biggest growth challenge right now? I'll say getting views and discoverability. And then I will go ahead and enter my subscriber range. And after entering this here, it says great context. So, it actually went and got the context. This tool is getting better at actually doing that. But what it's giving me here is pretty generic advice because I haven't given it enough context. And I didn't tell it exactly what I want, which is real video ideas. So, it's giving me ideas, but these are very surface level. This is the type of information that it would give to any productivity channel. But if I want to get specific information that's actually going to help me personally, I can start a new chat and we can use our bulletproof prompting sequence.

So, first I'll give it my goal. My goal is to create YouTube videos that will perform well for my channel. Then I'll go ahead and give it the context. I'll just say the analytics from my past videos are attached as context. Then I'll grab my analytics from YouTube. I'm just going to download them as a CSV. Then I can drag all of those analytics directly into Claude. So I've given it my goal. I've given it the context and now I have to give it the action. One of the most important parts. I say, "I want you to analyze my past video stats from my channel. Then research my niche." I'll add with web search and share three videos that would perform well for my channel based on current trends in the industry and in my niche and what has worked for me in the past. And finally, I need to give it the output format. So, I'll paste in the output format. Your final output should be a bullet list of those video ideas with title, length, target, and brief summary of what that video should cover.

Now, let's go ahead and send this off and compare it to how well it does compared to that generic response from before. Now, before I send this off, to get the most out of this, I've upgraded to the pro plan. So, let's go ahead and send it now. And now, it's coming back with that exact format that I wanted, right? remember my output format, a bullet list with the title, the length, target, and also a brief description of why this video would do well, and a bit of a summary. And the first video that it suggests is a Claude Code full beginner's guide, which is the video that I'm actually filming right now. So, that's actually pretty funny. It also recommended N8N plus Claude AI, build a complete business automation step by step, and seven AI tools you need in 2026. And it told me exactly why. And the nice thing about this is it's basing it on my past videos. It's not just giving me that generic canned information like the first chat that we started here.

Now, if you want to improve your prompting and you want to learn all of these other amazing AI tools that are coming down the pipeline right now, then I recommend checking out our courses on our website. I'll leave a link to this in the description below. This is where we really dive in and go deep. If you want lifetime access to any of these courses on our site, then you can buy them out right here and you can get them for a lifetime or you can join our community and get access to all of these courses. And you can also jump on with me and my brother Drake from AI Foundations on a weekly call. From there, we can get you and your team up to speed with AI. And if you want us to just build out the automations, deploy your apps, and do all of this done for you as a service, you can also apply below to work with us directly. I'll leave those links below.

Now that we have a few chats in our sidebar, I want to show you more how you can use in-depth search to find what you were looking for from a past chat. You can just hit the search button here, and then if you go and type out something like productive dude, you might not see results here because it's just looking at the titles right now. But if you want deeper search into your past chats, then you can hit go to settings to enable deeper search. And then now you can actually turn on search and reference chats. This allows Claude to search for relevant details in past chats. So I'll just go ahead and turn this on. Now it can search deeper when I go up here to click search and I type in a query. If you want a full screen view of this, then you can click on chats in the lefth hand sidebar and you can get a full page where you can search and you can see all of your chats here. These chats are great for just one-off questions or conversations that you want to have with Claude. But if you want more of an ongoing project, then what I would recommend is checking out the projects tab over here. And what you can do is you can click new project and you could create a project. And I could just call it productive dude for instance. And this could be a project that I can keep coming back to and adding to and really just forming and growing as I talk about my YouTube channel with Claude. And here it asks, what are you trying to achieve? So I could just say something like growing my YouTube channel and driving traffic to my offers. And then I could hit create project. And within a project, I can have siloed off memory within just this project. I can also add instructions for how I would like this project to work. And I can upload files that are relevant to this. So for the files, I can just drag in my stats over here. And I can also add some instructions. And it gives us some suggestions here for project instructions. But I can actually just use our GCAO prompting sequence inside of these instructions so that I don't have to type out the goal, the context, action, and output format every single time. I'll just say the goal of this project is to be a co-worker along with my YouTube channel and help me to develop a better channel. I want to grow in subscribers. I want to grow in views. But most importantly, I want to grow my business using this YouTube channel. A preference that matters to me is I don't care much for vanity metrics. Views and subscribers should simply be an indicator that I'm getting in front of more people that could be potential prospects. It's not something that I like to focus too heavily on. We use this as a signal that I'm doing something right, but we don't look at it as gospel. So, I've given it the goal and I've started to give it some context. Now, I'm just going to continue with the context. I'll say here's a link to my channel and I've given it the link now. And I'll also add there are files attached to this project as context and I'm going to continue adding files. If you need me to add more files so that you can learn more about the behind the scenes on my channel, then you can ask me to add those. Also, I want you to make sure that you use web search when you're working on this project to make sure you're getting updated information. And now for the A for action on my project instructions. I'm going to tell it to act as a consultant because this is going to be more general purpose. Act as a consultant. Give me all of the right questions that you need in order to point me in the right direction. And then finally, I'm going to give it output format. or your outputs. Make sure to match my energy and match the length of what I'm saying in the conversations. So, if I give you a short message, respond with a shorter message. If I ask for something in detail or I give you a long detailed prompt, give me a more detailed answer. I want you to talk back and forth with me in a conversational manner. So, now I've constructed these project instructions. So, these are going to be added to any chats that I have in this project. I'll go ahead and save the instructions. And now, since it has those instructions and these files, it's going to have that in the context every time I chat within this project. I'm just going to say, give me a 90day road map for my channel to grow it to the next level. And now, it's really talking to me like a consultant because it's saying, giving you a generic 90-day plan would be a waste of your time. Here's what I need to know. So, it's asking me questions just like I instructed it. What's the actual business attached to this channel? What are you selling? How does someone go from watching your video to becoming a paying customer? What does the funnel look like right now? Are you currently getting leads or sales from the channel? And if so, which videos are driving them? So, it's really asking me some hard-hitting questions thanks to those project instructions. And if I go back over to projects now, I can actually see these all organized. There's this initial one that it gives you on how to use Claude. Then there's also the productive dude project. And if I click into it, I can see all of my chats within there. And I can continue adding to my instructions, my files. And as this goes on, the memory is going to get stronger and stronger as well. And if you want to, you can even star your project so that it shows up over here in your sidebar and you can have quick, easy access to it. And where this gets super powerful is when you have a team of people working on whatever it is that you have your project around. With that, you can actually share the project and then start inviting other users that can interact with this project as well. That way, everyone's on the same page.

Next, I'll go over artifacts and visualizations so that you can see some of the amazing things that you can do just inside of Claude alone. You can create living, breathing documents and mini apps with artifacts. And to start things off here, I'll just go over to the left-hand sidebar and click on artifacts. And one of the cool things about artifacts is is they actually have all of these test artifacts that you can check out and play around with to begin with. They even have games in here. So, if I go to AI platformer game, I can show you exactly how this artifact works. We can customize the artifact or we can just start using it as if it's an app. So, I can hit start. I can select a theme for the level. And now it's generating that level. And boom, it pulls up a custom level for me where I can move around. I can jump around. I can grab stars. There's enemies around here. So, I can just go all over the map here and I can play the game.

Now, let's take a look at some more practical inspiration artifacts that they have. I'll just go back. And as you can see here, we have PRD to prototype. So, this one allows me to actually input a document and then it will turn that into a prototype of an app. The raw note transformer will allow me to upload written notes and turn those into an actual document. They have QR code generators, attachment style quizzes, bedtime story generators, and they even have this forest explorer where if I open this one up, I can be inside of a 3D world where I can move around and I can look at the forest. So, this is actually pretty cool and it's crazy that you can do all of this just inside of Claude's web app. And the opportunities are endless.

But, we haven't even created our own artifact yet. So, that's what we're going to be doing next. So, I'll go back and I can just hit new artifact and then it's going to walk me through and I can basically decide what I want to build. So, I can start from scratch or in this case I'm going to be creating an app or website. Let's imagine that I wanted to create a personal portfolio website where I share information about myself and some of the projects I'm working on and I give people an opportunity to work with me. So, let's go to apps and websites. And now it's going to start pulling me into a chat. And as you can see, that chat even shows up in our sidebar here. It's asking me what kind of app or website do I want to build. And I'm just going to go with the portfolio/personal site. Do you have a specific idea in mind? I'm going to say I have a rough idea. Then it says, what's the vibe or style you're going for? Let's say bold and modern. It looks like it has a couple more questions. So, let's just continue here. It's asking, "What field or profession is the portfolio for?" I'm just going to say freelancer/conultant. What sections do you want to include? This is a multi select, so I'll say about me. I'll say skills and contact. And then we'll go ahead and send that. And now it's creating a plan for my portfolio site. So I can read through the plan. And then I can confirm when I'm ready to start building. I'm just going to say yes, build it out. And now, as you can see, it opens up this sidebar right here where it's showing the code right now as the AI is actually writing that code for us. But when it's done writing this code, it's going to render the website for us. So we can see exactly what it looks like and we can ask for any edits. And boom, now it's done. And it's generated the first semblance of a website for us. So this actually looks pretty good, but it also looks quite generic. We have those generic AI gradients that it throws on the name and in these bars here. It's using emojis everywhere. I like the direction of this, but it's not quite personalized yet, and it just looks a bit generic.

Quickly, I wanted to show you that you can switch between the I up here, which shows you the actual rendered site, and the code here so that you can read the code if you want to, or you can copy it, or you can even click this drop down here and download as HTML so that you can hand this off to a developer or another AI tool like Claw Code. I'm going to go back to the visual view here, and what I'm going to ask it to do is to generate some different variations of this site that are a little bit more unique to me. So, I'll just say generate 10 styling variations of this site that are all unique. I want this to look like something that AI didn't generate, and I want it to feel like it has a unique style. So, give each style a name and generate 10 styles that I can look through with a tab switcher. Now, I'll send this off. And it goes on creating a separate artifact. If I wanted to go back to my previous one, I can just click right here, and it's saved for me. But I can see the progress as it goes through and writes this new artifact with the 10 unique styling variations. And as you can see here, it's creating different styles like this dark luxury style here. And it's going to apply all of these styles to that page. And I'll be able to click through and see the different styles. And now we have studio hand. It's generating another style for me. So it's on style number four. I'll be back when all of these are ready to show you. All right. So it's on the 10th style now and it's about to wrap up. But guys, look at this. Isn't this insane how many lines of code it's able to generate here in one shot alone? All right, it looks like it's almost done. And let's take a look at what it created. Boom. This looks a lot more unique and a lot more creative for my style. But this is just one of the styles that it provided. And the hovering works. Everything looks great. Now I can just click through these and decide which one I like. The layouts are just so much better on these than the original design that it gave me. Let's just click through and see what else they created. This dark luxury style looks pretty cool. We have studio hand. I like how it even integrated some different fonts. Neon terminal. Oo, this one's pretty cool. Editorial broad sheet. This one looks like a newspaper. We have retro Memphis. Very flashy. Soft organic glass and chrome. This is the original one that it gave me, but they just polished it up a bit. And then we have bold slab. And this one actually looks really nice as well.

Now, if I wanted to, I could publish this artifact. And I could publish and copy link. This allows me to share this link with anyone so that other people can see this artifact as well and they can utilize it. This is great if you're drafting up a website for a client or if you have a presentation to share. You want to share some ideas and you just want people to be able to click around. I've used this in the past when doing web development to share different font styles or different website layouts with my clients before I build the full-fledged website. But website development isn't the only thing you can do with an artifact. You can also create helpful mini apps that you can come back to and use time and time again. Let's imagine that I'm jumping on a sales call with someone who's in a different time zone. I could quickly tell Claude, I'm jumping on a call with a client from Saudi Arabia and I am in the Eastern time zone and I would like you to create a slider app.

In an artifact in the bold slab style that I can use to simply drag and see what time it is in Eastern time versus what time it is in Saudi Arabia when I'm scheduling our next call. And I can send that off and it just gets right to work. And would you look at that? It gave us the exact style that we wanted and an interactive slider that's going to change depending on the time. So, if I have a meeting that I want to schedule for 10:00 a.m., I know that that's 5:00 p.m. in Saudi Arabia for Arabia standard time. And it even tells me down here if this is a good time. So, as I drag the slider, I can find a time that's within both of our work hours so that I can be respectful to the client. For instance, if I wanted to have a meeting at 2:00 p.m., this would be 9:00 p.m. in Saudi Arabia. That's so it says evening. And if I go way too crazy with this, it says outside of work hours. This could be a helpful tool that you spin up in a minute and then you have it open right before you jump on the Zoom call with the client.

Next, I want to talk about inline visualizations. And this can be used in a project or just in a regular chat. To showcase inline visualization, I'm going to be using my productive dude project that I already created for my YouTube channel. I want to create a nice view of some different charts and some analytics so that I can really visualize my YouTube channel's path. So, I'll open up our sidebar here and I'll click into my starred productive dude project. Then, I'll just tell it, I want you to create an inline visualization of the analytics from my channel. I also want you to include some future projections for my channel and some best cases, middle-of the road cases, and worst case scenarios. And I want you to tell me what I need to do to hit the best case versus the middle of the road versus the worst case. And I'm going to switch this to opus 4.6 so I can get the best response. Now I'll go ahead and send this one off. So it's reading the files that it has available to it. I can even click in here and I can see exactly what it's processing and what it's going through. And as you can see here, I can click into the different CSVs and I can see exactly what it was reading and I can view it in here, similar to how I can view the artifacts. But for now, I'm just going to close this and we'll see if it generates that inline visualization. And boom. As you can see, it is now coming to life. We have all of the data populating in. And as you can see, this is inlined. This isn't a separate artifact. It's just a natural way to chat with your data. And right now, it's modeling this out and polishing the dashboard. So, we'll see what it generates for us. Boom. Would you look at that? It came through with all of the data now. And it even mapped out the projections. And I can see exactly what's possible with my channel if I go down these different roads. It talks about what would lead to the best case scenario, what would get me the middle of the road and what would be a worst case scenario. And I can see monthly view trends here. I can see daily views over time. And it gives me some totals up top here. Now, if I really like an inline visualization, I can hover the visualization and up here, I can copy it to my clipboard. I can download the file or I can actually save it as an artifact and turn it into an artifact that we can then publish, share, or just save for the future. So, let's go ahead and save this as an artifact. And now it's rewriting it as an artifact so that we can save it for later. While it's working on that, I can also go back to my artifacts. And then I can hit your artifacts and I can see all of the artifacts that have created so far. So this kind of acts as your app dashboard for all of your prototypes and mini apps that you've built. And you can always come back to them and use them whenever you want just by clicking on them.

Next, I'll show you how to get the most out of Claude using Claude Desktop, which gives us access to Claude Co-work, Claude Code, and just the normal Claude that we've been using. I'll just click the download button in the lower left hand corner here. And this is going to show me all of the things that you can install. The first thing is Microsoft Office, which allows you to analyze data using something like Excel or build presentations in PowerPoint. But what we're looking for is this co-work tab right here. So, I'm just going to download for MacOSS since that's what I'm on. So, I'll just hit download and then I'll open up the install file and I'll drag Claude into my applications folder. Once it's saved, I can just double click on Claude and hit open. Now I'll hit get started. And I'm just going to sign in to the same account that I was on with the web version. And now Claude is going to open with Co-work. So as you can see here, we have a few different options. We have dispatch and scheduled now. And we also have this ideas tab. You can see that we also have at the very top of the page a switch between chat, which is basically like our normal chat screen. We have co-work and we have code. So these are the three different modes of Claude that you can use and I'm going to be covering all of them. We've already discussed chat and as you can see here I have access to all of my conversations and projects that I built out in the web app earlier. I can go over to co-work and there are some additional options and then I can go to clawed code and there's even more options. I'm going to start with co-work and I'm going to dismiss these starter tasks because I'm going to show you guys how to set everything up and then I will dismiss this task list. And let's start by looking at some of the additional settings that you can edit on your actual desktop device. So, if I go down to the left hand side and I click on my name, I can click settings now. And in here, you're going to notice that when I go to co-work, it's going to actually have some additional options. Now, it shows that we can turn dispatch on or off, and we'll talk more about that later. It shows how we can add global instructions for our co-work sessions. These are preferences, conventions, or context that Claude should always know when it's working on your device directly. We have Claude in Chrome, which allows Claude to actually use Google Chrome as a browser to click around and look around websites. We'll also touch on this in a little bit. And then we have this new option down here, desktop app. So, if I go to general, there are some settings that we can change in here. We can choose to run Claude on startup. That means when our computer first turns on, if we want Claude to turn on with it, we can flip this switch right here, and Claude's just going to automatically start up when we turn our computer on. We also have the quick access shortcut, which is set to tap option twice on mine right now. Or we could use option plus space or set up a custom shortcut. I'll show you how this works really quick. So, I can be anywhere on my computer and I can just double tap option and it's going to pull up the Claude bar down here. So, then I can just start a new chat or I can click into a chat that I've already created and I can continue chatting with Claude. And while this is open, I can also drag to screenshot something and it will automatically add that as an image inside of Claude here. Then I could say something like, "What is it that I'm looking at?" and send that off and it's going to open up that chat where I pasted those screenshots. So, this can be really helpful if you're clicking around your computer and you have a quick question that you want to ask Claude. Let's head back over to the settings and we'll go ahead and dive a little bit deeper. So, we've already touched on a few of these general options here. You can also choose to show Claude in your menu bar if you want to. And this one is super important. If you want Claude to work with dispatch, you can turn on keep computer awake and that will keep your computer on so that Claude can continue using it even when you leave the office. That way later you can chat with it from your phone using dispatch. And then if we scroll down a bit further, we have the browser use settings. We can allow all browser actions. If we don't want to approve every time that it asks to do something in the browser, we can just turn this on. I'm going to leave it off for now just to be safe. And then we have computer use. If we want Cloud to be able to actually control our computer itself, take screenshots, move our keyboard around, move our mouse around, it's able to do that if we turn this setting on. And then since it hides the app in the background when Claude is using it, we can actually turn this setting on if we want the app to reappear after Claude stops working on it. And just to be safe, you can also deny apps right here by adding them to the denied list. If you haven't already, make sure to give Claude accessibility and screen recording permissions on your device. And then if we go to extensions here in the sidebar, we can add extensions. So you can browse existing extensions. And these are just any type of connector that you can use on your device. This is going to be even more powerful than the connectors that I was showing you in the web app because some of these can even do things like read and send iME messages. And since I'm on a MacBook and I have iMessage installed, it can do all of that for me if I enable this connector right here. Or for instance, I can give it this file system access and it can access, read, and write files on my computer itself. I'm going to close out of the extensions here. And if I click on advanced settings now, you can enable auto updates for extensions if the extension gets an update and you don't want to have to go and renew it. I'll just turn that on. And it detects what I have installed on my machine. Node.js and Python. If you're an extension developer, you can actually install extensions here, install unpacked extensions, open extensions folder, and open extension settings folder here. So, this is just a nice developer tool. It's a way for developers to access and test out their extensions after they've built them. And then of course there is the developer tab as well where you can load in local MCP servers which are similar to extensions or connectors.

Now one of the first things that I do when I open up co-work is I get everything organized on my computer. I don't drag the files around my desktop or my downloads anymore. I just have Claude do it for me. So if I hit new task up here while I'm on co-work, I'm able to actually tell it to do that. So, I'll just say, I want you to read my downloads folder and see all of the files that I have there. Then, I want you to move them into my documents directory. I have a bunch of subfolders inside of documents for things like photos, templates, videos, etc. So, I want you to move those files that are in downloads into those specific document folders. All right. So, I'll go ahead and send this off. And now, in co-work, it actually generates a task. So, you can see the task's progress here. If it's a longer task, it will actually generate a checklist right here. You can see the working folder by opening this up right here. And if it's in a folder at that time, it's going to show you what folder that we're working within. So, right now, it's loading tools and it's requesting access. So, first and foremost here, I need to give it access to the finder. And I also need to give it script editor access so that it can write a script to organize these files for me. So, I'll go ahead and allow for this session. And now Claude is actually opening up the script editor for me and it's using my computer here. As you can see, it says Claude is using your computer. And over on my other screen here, it has it outlined in this orange glow. And I can see it actually doing things on my device. It's opening the folders. It's moving things around. It's accessing the script editor. This is what my other monitor looks like as Claude is processing this. Co-work is still running, so I'm going to let it finish up this task. So, it just finished cleaning up most of my downloads folder, but there were still a few file types that it didn't pick up. So, now it's going to sort the HEIC photos and theOV videos, and then it will be complete. And now, as you can see, our chat has gotten quite long. So, Claude is compacting the conversation. This basically just summarizes everything that it's done so far, and it doesn't have to keep passing all of that information from the conversation. It can just move forward with that summary. All right. So, the good news is this finally finished and it moved all of my files out of downloads into the specific documents folders. So, it moved all those HIC iPhone photos from downloads to the documents/phos. Everything got moved to the correct folder. The bad news is I'm running low on usage. So, what I'm going to do is I'm actually going to switch over to my main claude account for the rest of this tutorial.

Next, let's take a look at scheduled tasks. So, if you go to scheduled in the sidebar here, you can then create a new scheduled task. Now, make sure that your computer is kept awake and you have this setting on or you can turn it on in your settings like I showed you earlier here. If I wanted to schedule a task to take place at a certain time, I could just hit new task. So, let's say that I wanted it to clean my desktop and my downloads folder once a day. So, I could just go to name and I'll just call this clean files. And I'm just going to say for the description, cleans my desktop and downloads folder once per day. And then for the prompt, I'm actually going to write out my entire prompt. So I wrote out the prompt here. Check my desktop and downloads folder. If there are files in there, I want you to move them to the documents subfolders. Inside of documents, there are folders for things like videos, photos, PDFs, etc. If there are any files on my desktop or in downloads, I want you to move them to my documents subdirectories. And then I can set the frequency here. So, let's just say I want this to run daily. And I'll set the first task to run at 8:00 p.m. And then I'll go ahead and save this. Now, at around 8:00 p.m. every day, this task is going to run. And right now, it's scheduled for today at 8:07 p.m. I can also click on this task, and I can run it manually right now, or I can edit this task to change any of the settings that I have here. Let's go ahead and run this to see what it looks like. All right. So, as you can see, it pulled it in here today at 8:00 p.m. and it says running. So, if I click on it, I can actually view that process happening, but normally this could happen in the background. So, now, as you can see, Claude is active on my desktop, and hopefully it can help me start cleaning up some of these files. All right. And just like that, it's gone ahead and completed the desktop cleaning. So, now I don't have a messy desktop. But it also looks like it moved the path for the mask around my camera here on OBS. So now I guess I'm in a square.

Next, let's take a look at projects because they're a bit different on co-work than they are in normal chat mode. So if I go over to projects on the sidebar, you'll see that I don't have any projects in here because I haven't created any for co-work. If I click new project, it's going to work a bit differently. It's going to ask me, do you want to start from scratch, import a project from chat? So, I can actually just click this and open up a project that I have over in chat or use an existing folder. So, I'm just going to say start from scratch. And what this does is it actually creates a projects directory within claude. And this allows me to basically create a folder and tie it to that project. So, a folder on your computer contains different files, different documents, and then other subfolders within that. And this is what acts as your project for Claude Co-work. So, I'm going to create a project called prospect research. And then for the instructions, let's imagine that I want to research and enrich a document of prospects here. So, I'm just going to say this project is meant to use web search and save files about prospects in the form of CSV and PDFs so that I can generate proposals and mockups for websites for local companies. And then if I wanted to, I could add files here, but I'm just going to have Co-work actually create the files for me. So, I'm just going to click create. And now we're in the prospect research project. I can see it here as well underneath the chat bar. So, I just said, "I want you to do a web search and find local companies in Southwest Michigan that have a website that needs to be redesigned. Find the 10 most promising prospects and save them to a CSV file in this project." Now, I'll say, "Let's go." And I see that this chat is kicking off. And as you can see here, this task has been paused and it needs a response for me. So, I'm going to click respond. And what it's asking here is what it should focus on in terms of area. I'm just going to say all of Southwest Michigan. I'll say home and trade services and professional services as the categories. And then this is a great question that it asks. What signals count as needs a redesign? I'm going to say not mobile friendly and thin content/no SEO. And then I'll send that off. And now this is going to be a lot more tailored when it does its web search. And over here in prospect research, I can actually click on this folder right here to open up that folder and see what it's creating inside. And as you can see there, it just popped that CSV file in. So, it must have found some prospects that I can actually reach out to. So, let's go ahead and open this up. And just like that, it created that CSV file for me. And it's right here on my device. It also included a line on why this is promising. So, it shared exactly why we might want to reach out to these companies. It also shared the source of how it found them. So, I can take all of this information and I can actually reach out to these companies. Now, now I'm going to get it to research these companies deeper, verify any problems with their website, and take a screenshot of each one of the homepages on their site. Then I'll have it draft a cold email that I can send to these companies. So now I'm going to say, "Now I want you to go and take a screenshot of each one of their websites and include them in a subfolder within this project. I also want you to draft a markdown file with cold emails to each one of these companies. I want you to include the images inside of those write-ups that you do. And I want you to make the cold email short, punchy, and include relevant details about them. Also, find the email for the primary contact for each one of these companies and include that as well. Now, I'll go ahead and send that off. Now, it says, could you open Chrome and make sure the Claude and Chrome extension is signed in and active. Once it's connected, I'll navigate to each of the 10 sites and blah blah blah. So, I need to actually open up Chrome and make sure that I have the extension installed. To do this, I'll just go down to the left hand corner. I'll click on settings and I'll go to Claude and Chrome and then I'll make sure to allow all extensions. Now I'll open up Google Chrome and I'll head over and add Claude to Chrome. This is a Chrome extension from Anthropic. Just make sure that it says verified fromclude.com one extension and it should just say Claude. So I'll hit add to Chrome and add extension. It will ask me to verify and I'll just continue with Google. And once I'm signed in with Google, I can now hit authorize. And now I can hide the browser and I can go back and say, okay, Claude and Chrome is enabled now. Let's go ahead and continue with the task. And when I look at my context down here, I can actually see that Claude and Chrome is active. And I can see a preview of the website that it's on. So now it's grabbed the screenshots. And next, it's drafting up those cold emails with the email included that it found from the websites. All right. And boom, just like that, it's gone ahead and completed that task. And it's delivering the final files here. So now let's go ahead and check out the folder to see what it generated. If I open it up, I can now see that it has the cold emails.md. I've gone ahead and opened this with Visual Studio Code. And now I can see all of this markdown that it generated here. For the names that it could find, it actually included their name in the cold email for the decision maker. But for the ones where it didn't find the email, it included their phone number and recommended that I call them. And it got really clever and customized with these subjects. It says, "For this one, Matt, the $20 off banner is doing the heavy lifting." And then it says, "Your $20 off plus Google review combo is smart, but it's competing with a stock photo of feet in Flood Water that's pushing your services and trust signals below the fold." And then it just goes on to give them some good feedback. So, I could actually take these emails, send them off to the people where it actually provided an email, or I could call the people who didn't have an email provided, and I could propose a website to them. And then I could actually build the website using Claude as well.

Next, let's take a look at dispatch. I'll click on dispatch in the sidebar here. And what you'll notice is that by default, dispatch is turned off. You have to configure it. So, what you have to do is go to settings and then turn dispatch on. Then, if I head back over to dispatch, you'll see that it may connect to your phone app. I have the app already installed on my phone and I'm able to chat with it from here. You may have to actually sync it up with the clawed app on your phone, but once you do, you can just type messages from your phone, and as long as your computer's on here, dispatch can do all the heavy lifting. So, I'll just say, "Hey, what projects do I have from my phone?" And I'll send that off. And as you can see, it came through there.

Next, I'm going to show you how to create connections to outside apps using Claude Co-work. This is very similar for clawed chat, but as I've said before, co-work is just a bit more powerful because has access to your computer and extra connections and skills. But I'll just click the plus button here, and then I'll go down to connectors. Now, I have some connectors already in here, but I'm just going to add a new connector as a test. And by default, as you can see, there's all of these connectors, and you can actually search for a connector as well. So, I'm just going to search for Notion because I want to connect to a CRM that I have in Notion. So, I'll hit the plus button on notion, and it's going to open up my Notion account here. So, I'll just go ahead and sign into Notion with Google, and it says connect with Notion MCP. And I can just hit continue here. And now it says connected to Notion. So, now I'm connected to Notion. And here I can change the tool permissions for Notion. So, inside of here, I can read what the different tools are, and I can see what's always allowed versus what needs my approval versus what I don't want at all. So, I can just completely turn it off. So, if I don't want Notion to be able to create comments through Claude, I can just turn that off here. And I can just go through here and set my permissions. Now, I'll back out of customize here, and I'll click the plus button, go to connectors, and I'll make sure that notion is turned on. I'm going to turn off these other tools here for right now. And over here in notion, I have my Apex Consulting CRM. And this includes all of my contacts for a fictional consulting company, but this could be your exact CRM if you wanted to. And you could have all of this rich data on these prospects. And you could even have things in here like notes so that you can learn more about them. So now I'm just going to tell Co-work to use my notion connection to find my Apex Consulting CRM. And then I'll go ahead and say, "Let's go." So it found it here. And now I've just asked it to give me a summary of what's inside. So now it's actually giving me a full schema of what this database has and what levels that different people are at. So I could already start gaining insights just by asking it to tell me a summary of what's in the CRM. In here I have last connected date. So I could actually ask it to give me some people that I could follow up with that I haven't talked to in a while. So I've just said give me a list of people I should follow up with that I haven't talked to in a while. I'm going to send that off. And now it's giving me a list of all of these people. And because it has the email in here, it could actually draft up an email to send these people from my contacts. So, let's go ahead and also connect Gmail. I'm going to add a connector. And I'll just type in Gmail. And then I'll go ahead and connect it and hit continue. So, now I'm connected to Gmail and Notion at the same time. And since my notion CRM has those emails in there and all of the information about the prospect and who I haven't followed up with in a while, we can now send emails to those prospects that I haven't talked to in a long time. So, I'm just going to say, go ahead and draft an email using the Gmail connector. Don't send these off. I just want you to draft them for everyone that I haven't talked to in a while with a relevant follow-up. I'll go ahead and send this off. And now, as you can see, it has Gmail in the context here. and it's creating the Gmail drafts. And here it's asking me for permission. I'm just going to always allow for this task. And it's creating some draft emails. All right. So now it's gone ahead and drafted 15 emails for me in a matter of minutes. I'm going to go ahead and click on this William Chang email. And we'll go ahead and open it up. And as you can see here, it drafted up this beautiful email to William. It says, "Hope tax season treated you well this year. When we last spoke, you mentioned circling back in May once things calmed down. I wanted to get on your radar a little early so we can find a good time that works. This is perfect. And then I say, given the size of Chang Financial and the questions you raised on ROI and implementation timeline, I think the one-on-one intensive is still the right shape for what you're trying to build. Happy to walk you through what the first 30 days would look like, specifically for a 15 person advisory plus accounting firm. Would a quick 20inut call in the next two weeks work for you? Best Carter. Perfect. It wrote this email out about exactly how I would send it. The only thing is it included this mdash here, which doesn't look natural. And I could have told it to not include that, but I'll just put a comma here instead. And now this looks perfect to go ahead and send. And now in notion, I'm going to go ahead and search for William. And I see William Chang here. I'm going to go ahead and open him up. And now I can see that it hit the details spoton. He does have a 15 person financial advisory and accounting firm. and he did say to check back in in May. So Claude got the details right on the money. And it did this for all of the consulting clients. So imagine what this could do for your CRM if you had it in something like Notion or even HubSpot. Pair that with Gmail and you can send out a bunch of different draft follow-ups to potential prospects.

Next, we're diving into skills. More specifically, how to create and use skills inside of Claude. You can create skills inside of the Claude web version, and you can even upload them there. But again, they're not as powerful as when you do this inside of Co-work because Co-work has direct access to your computer. It can run scripts and overall it has many more tools that are available to it when it's executing those skills and creating them for you. What is a skill? Well, a skill is basically a repeatable process that you can have Claw do the same way every single time. It goes through that step-by-step workflow, uses tools that you've already defined, and it creates a reliable workflow that you can always trust so that you're not just going back into the chat and it's doing it differently every single time. With a skill, you get the same guard rails, the same process, the same tools, everything, and you get it perfectly consistent. Now, there are three ways to create a skill. Number one is to just simply describe it to Claude. So, you can tell Claude what you want automated. It'll ask you questions, and it will build the skill for you, and then it can also run the tests for you. So you can give the skill an input. You can tell it to execute that skill. And this is best for new ideas. The second way is to reverse engineer something that Claude has already done. So you get the output that you want first by just prompting normally, going through a process, going through a normal conversation with Claude, and then you say something as simple as make a skill that does exactly what we just did. And this is best for if you have proven results and you got a good result with Claude just chatting with it normally and you want to turn that into a repeatable process. And then the third way to create a skill is to actually upload a skill that someone else has already created. If you want to create powerful skills, then I recommend getting our free resource, the DBS framework skill. My brother Drake created this and I'm going to explain exactly why it's the best way to create skills inside of Claude. What is the DBS framework? What is this skill? And why does it matter to you? The D in DBS stands for direction. And these are the instructions that live in the skill.md. This is the name and the description and the step-by-step workflow of the skill. And you also sprinkle in rules and guidelines here. Then we have B, which stands for blueprints. These are examples, references, and guidance that's going to help tailor your outputs. Think of these as the relevant resources that are going to make your skill more powerful. Finally, my favorite part is solutions. And this makes your skills more kinetic. And you can think of this as computer programs that can live inside of your skill and be used again and again. If there's something inside of a skill that Claude can't normally do well on its own, you can actually have Claude help you create a script or a bit of code that it can use to enhance that process and get to your result faster and more accurately. And this is the DBS framework. I'll leave a link to our post where you can actually access this skill file. Once you get down to this point in the page, you're going to see the DBS framework skill.zip. You can just hit download here. Once you download that file, you can go to customize, click on skills, and then just click the plus button right here and create a skill, and just hit upload a skill. Then you can simply drag in that file that you just downloaded from our website into this area here, and it's going to upload that skill. And as you can see here, you now have the DBS framework skill. You can see exactly how we set this up for helping you to create better skills. And in my opinion, this is the best way to create skills outside of the normal skill creator that Claude gives you out of the box.

So now if I go to new task in co-work, I can click the plus button right here, go to skills, and then I can click on DBS framework. And now it's going to put that skill into the context of my conversation. And I'm just going to create a very basic skill to start out. This is the bottom line, simplest way to create a skill. You just pop in DBS framework here. And for this skill, I want to create a recipe generator so that I can more efficiently use the ingredients that I have in the kitchen. So, I'm going to tell Claude exactly what I wanted to build here after I've put in the DBS framework. So, I just say, "So, I want you to build a skill that walks me through a Q&A and questionnaire process to find out exactly what I have in my kitchen. Once you figure out what I have in my refrigerator, what I have in my spice cabinet, and what I have in my pantry, you're going to generate three recipes based upon that. The skill should be allowed to use web search, and the skill is invoked or started basically by typing in /cook. Then, it begins the questionnaire process." So, I'm going to send this off along with the DBS framework skill selected. And this is going to create that skill for us based on the merits of the DBS framework. And then I'm going to show you how to create some more advanced skills, some more helpful use cases. But this is just a great way to kick things off and show you how to create your first skill. It's very low stakes and you could actually use something like this to help you cook dinner tonight and be more efficient with the ingredients that you have. So, it's creating the skill.md now. So, it's giving Claude the future instructions that it will need to go through this process. And it's giving it the step-by-step workflow based on how the DBS framework says to create these skills. And now it's saying Claude wants to use creating a file. So, I'm going to always allow this. And now it's created the skill. So, I'm just going to hit save skill. Now that I've created the skill, let's go ahead and give it a try. I can just type in /cook. And look at that. It's right there. So, I can just click on cook and then just say let's go. and it's going to begin walking me through a structured Q&A process. So, as you can see, it even pulls up the Q&A right here in Claude because we're using the user question and answer tool. So, it says, "What proteins and dairy do you have in the fridge right now?" So, I can say chicken, I have eggs, I have cheese, butter, milk. Then I'll send that. What produce and condiments are in the fridge? I'll say fresh herbs, lemons/limes, onion, garlic, carrots, celery. Let's go ahead and send that. And next, it's asking what I have in my spice cabinet. And if these answers don't line up, I can just write something else down here. So, let's just say, let's imagine that I just have salt, pepper, garlic powder, and paprika. I could just write out the exact list of what I have in my spice cabinet right now, and then send that off. And finally, it's asking, "What grains, canned goods, and oils do you have in the pantry?" I'll just say, "I have avocado oil, nay pasta, olive oil, canned tomatoes, and rice." All right, let's send that off. For dietary restrictions and allergies, I'm going to say none. How many people just for two here? How much time do you want to spend? I'll say under 30 minutes. I'm too busy and clawed co-work. Cuisine mood, let's just say surprise me. And now it's going to give me three different dinners. And this was an entirely customized process that we can run back again and again because we've created the cook skill. It recommends lemon garlic chicken pane. It recommends Spanish style chicken and rice. So, it's going all over the map here. And then as I'm reading through here, it just pulls up which one of these are you making and do you want extra help. So it's going to be able to go with me through the process. And you can even bring this skill onto your phone later down the line. Let's go with the lemon garlic chicken pane walkthrough. And perfect. It's giving me a more detailed list of instructions and walking me through exactly how I can do this. And what we've covered here so far is only the tip of the iceberg when it comes to skills.

Now, I'm going to show you how to combine skills and connectors so that we can do repeatable processes that integrate with all of our business operations and all of our apps. Next, I'll show you how to create a more advanced and strategic skill that allows you to use connectors so that you can create repeat processes and automations. First, I'll load up the DBS framework just by typing DBS and then hitting enter. And now we have the DBS framework active. And then I just say for my prompt, I want to create a skill that looks at my Google calendar every day using the Google calendar connector. And I'll show you how to set that up in a moment. It tells me what's on my schedule for the day, giving me a daily brief when I type in /aily. So that's what's going to invoke this skill. I want this daily brief to be formatted as action items. I also want to look ahead at the rest of my week and have the task suggest things I should be doing to prep for later in the week. And I'm just going to add a new connector quickly so that it can see what that looks like. I'll go ahead and hit manage connectors and I'll click on my Google calendar down here and hit connect. It's going to ask me what I want it to be allowed to do and I can just select all if I want it to be able to do everything that it is asking for like viewing my calendar, downloading my calendar, seeing, editing, and sharing things in my calendar or even permanently deleting calendars. So, the permissions are pretty wide open here. But if I hit continue, what it's going to do is it's going to allow me to click on Google calendar here. It's going to allow me to go through and decide what events that I want to allow. So, let's just say that we don't want to be able to have it delete calendar events. I also don't want it updating calendar events, but it can create new ones. Now, I'll back to co-work and my prompt is still here. I'm just going to click and we'll hover over connectors again. And you'll see now that we have Google calendar enabled. So, let's go ahead and send off this skill creator with the DBS framework. So, now it's walking through the steps of the DBS framework. So step one is to define the task. This will be a bit more of a back and forth process. It's asking me some extra questions here. So Google calendar connector is the first question. Do you have the Google calendar MCP connected in co-work? If not, we'll need to set that up first since the skill depends on it. Yes. So I'm just going to say number one. Yes, it is connected. And then number two, prep suggestions depth. When you say suggest things I should be doing for prep, are you thinking practical logistics like prepare slides for Thursday's presentation or review agenda or do you also want it to flag things like travel time conflicts? I'm just going to say both and we'll send that off. And now if I scroll down here, it is asking to run this command right here for this session that it's created for daily brief. I'm just going to allow. And the DBS framework is doing a lot of the heavy lifting here. It knows how to create the skills based upon that. It ran a brief test. It looks good. I'm just going to save the skill and then I'll go to new task and I'm just going to type in /daily and I'll show you exactly what this looks like. So, let's go ahead and send it off. So, first it gets the day and it says here that it used Google calendar integration and you have one event today, a morning errand. Then the rest of the day is wide open from 10:30 a.m. onward. So today's action item is just to drop off my wife and son. And it says after the drop off, you're completely free for 10:30. Now, let's see what it recommends for the week ahead prep. So, Friday I have the content strategy call with our content team at 10 to 10:30 on Discord with four attendees. Come with any content updates, ideas, or blockers to raise with the team. If there's a running content calendar or backlog, skim it before the call so you're up to speed. And then the rest of Friday looks open as well, so it's a good day to execute on whatever comes about.

So far, this tutorial has probably been pretty simple to follow. But we're about to get into the more technical side of Claude, which is Claude code. And while I believe that this is the most powerful thing that Anthropic has to offer with Claude, this section of the video comes with a bit of a caveat. It's going to get more technical. So, if you're a nerd like me and you have a background in this stuff, you're probably going to understand everything that I touch on with Claude Code. But if you're a bit less technical and there are some terms that could go over your head, then in that case, this next section of the tutorial may require a bit of extra learning outside of just this video. So, there are two ways that I'm going to recommend picking up on the terms and the methodologies that I'm going to be covering for the remainder of this tutorial. Option number one is to simply use Claude and ask it questions about the things that I'm saying in this video. It's going to do a great job of breaking it down and explaining it to you at a beginner level. The truth is, when I first got into this stuff, I had no idea what I was doing. And asking Claude helped me along the way. But what helped me even more is having a mentor that I could ask questions to who was miles ahead of me. So, if you need a mentor or if you want a mentor to shortcut the process a bit, then I recommend joining our community over at AI Foundations. You'll get access to all of our courses, including Claude Code Mastery. when you join. And in Claude Code Mastery, I really break this stuff down and assume that somebody's coming to me as a total beginner wanting to learn about Claude Code. But I don't just cover Claude in this course. I walk you through the technical foundations that you need to go into this confident in understanding the terminology and the methods that we use. And when you join AI Foundations, you also get access to other people who are doing this.

as well. Most importantly, you can join weekly Q&A calls where you can ask me and my brother Drake questions directly. And we've shipped and sold software using Claude Code. So any question that comes up for you, we spent 10 months actually building our software out. It wasn't just an overnight thing that we launched. We went through the trials, the tribulations, and we figured out exactly how to do it.

So, if you start to feel a bit uneasy during the rest of this tutorial or you start to feel like this is a little bit out of your ballpark, then that's when I would recommend joining AI foundations or just finding a mentor, any mentor that can help you through this.

Next, let's take a look at Claude Code. Now, there are two ways that you can run Claude Code, and they're actually very different from each other. There's running Claude Code in the terminal, and then there's running Claude Code with this nice interface that Claude gives us.

So the easiest way to run quad code is just to click in the top center on code here and then you're going to get this beautiful screen here where you can start asking it to do coding tasks on your computer. Now the first thing to take a look at is where it says local in the bottom right hand corner here. This is referring to what machine that you're actually using to interact with claude. So they have local here and you can go into the settings of local and you can set up your local environment with your environment variables. Think about this as the secret passwords that you have to access different accounts and things like that, but just in a developer ccentric way.

So, I'm going to close this for now. But if I go back into local, you'll notice that we can also add an SSH connection. And an SSH connection allows us to connect to a different device. So, if I hit add SSH connection, I can input all of the settings for an SSH connection. We're not going to be doing this today because I'm just going to be using my local machine. But then there's another method for connecting here and that is to add remote control. So if I hit add remote control, what this does is it allows me to run this command in any folder and I can basically access that folder similar to dispatch on my phone in the web browser from a different machine. I can basically use this computer with this command to access any folder in a sandboxed folder where you can actually use claude from anywhere while you're on the go.

Now, the last option here is to add a cloud environment. And this is basically going to allow you to use Anthropics servers, the creators of Claude, to run coding tasks on a GitHub repository for you. So, you can configure a new environment by hitting add environment or you can simply use their default environment here. And then you would just select a GitHub repository and it would create a cloud environment where it spins up that GitHub repository and you can do coding tasks on that codebase. It doesn't require you to have a virtual private server or your computer because Anthropic is going to run it on their machine and you can make changes to that code through Anthropic's servers.

Now I'm going to click back into here and we're just going to stick with local for now. Then we also have some additional model options here. If I click on the model dropdown, I can see that I also have Opus 4.6 1 million. And this is just a bigger context window. It's able to see bigger code bases and it's a lot more powerful than the regular Opus 4.6. You might also see new models cropping up in here. The models are probably going to get a lot more powerful over the coming months. But then there's also the effort tab down here. And effort is basically how long you want it to spend on a task, how many agents you want it to spin up to work on a task. So if you have a crazy task, like you want it to analyze your entire codebase or something like that, then you're going to want to crank this up to max. But if you're just doing a quick change, then you can leave the effort a bit lower here. It defaults to medium. And I have it on Opus 4.61 million.

Next is the permissions tab. So, if I click on this here, this allows me to auto accept edits, ask permission, switch to plan mode, or bypass permissions, which you can enable in the clawed code settings. This allows it to run pretty much any command that it wants to run, which can be a bit more dangerous, but it also doesn't require your feedback. It can just run. So, we'll click out of this now. And next, if I go to more options here, you'll see that we can add files or photos. So, this is great for adding screenshots or different things for clawed code to be able to visually see what it's working on. We also have a ton of different slash commands that we can use in claude code like batch. And then we also have unique features like loop which allows claude to run in a cron loop meaning you can have it do a task and run it every 5 minutes automatically. And then also similarly to loop we also have schedule. So you can create a scheduled task that runs at a certain time.

If I click out of here there are some more options that we can look at. We also have the connectors which are similar to the other connectors that we've talked about so far, just a lot more powerful because Claude Code is using them. And we have plugins. So, front-end design is automatically installed here. But if I go to manage plugins, I can see some more of the plugins that Claude Code is allowing us to use. So, if I click on the add plugin button and browse plugins, I can see a bunch of popular plugins that Anthropic has released. And plugins are basically combinations of skills and connectors. They allow developers to come in like Apollo to develop a Claude plugin that helps Claude use it better. So, it's a combination of markdown files, maybe code, maybe connections that are going to allow you to say use Apollo better or use Slack better or view PDFs, all kinds of different crazy things that you can do with these. And you can also custom develop your own plugins if you want it to do something specific.

Now, most importantly, let's go ahead and select a folder. And this is going to allow us to select a code base that we already have or choose a different folder. Now, by default, it's not going to allow you to send a message until you select a folder or a project that you want to work on. So, I'll click select a folder, choose a different folder, and I can just select my root directory and hit open. So, now it's in my root directory. Now, what I recommend when you're working on projects in cloud code is having a separate directory called projects or apps or something like that. I'm just going to call this folder, in this case, tutorial. And I can actually have Claude generate that folder for me. Create a folder called tutorial. When I hit enter, it says trust this workspace. And it just makes sure that I'm allowing it to do this. So I'll just say trust the workspace. And then claw code will begin spinning up this task. It's running commands here. It's running a command here to make that directory for tutorial. And now it's created that folder. So now if I start a new session, I can go to select a folder and I can choose a different folder. And you'll see here that it created that file called tutorial. So I can open that up and then this folder can act as my directory for all of the apps that I'm building with claude code.

Now if I want to create instructions for claude to know for this project, I can have it create a claude.md file in here. Create a claude.md file that just basically says that this is the directory for my projects for a tutorial that I'm creating. And I'm going to have subdirectories in here for all of the projects that I have. So basically, you can think of this as the root folder that is going to contain all of my projects that I'm going to be sharing with you throughout this tutorial. So I'm going to send this off and it's going to know to create that claude MD file in tutorial and it's basically just going to give it a brief description of the subdirectory. So now with claude code, I can even say things like open that file in Finder for me and it will open up that claw.md file. I'll just allow it. And boom. Right here it opened up that Claude MD file that it created. And it's a very simple file, but it just says tutorial projects. This is the root directory for tutorial projects. Very simple, but that's just an idea of how you can get Claude to quickly create files. And I feel that this is a lot quicker than using co-work.

Now, let's say that I wanted to launch and deploy that website that I created earlier as an artifact. That's exactly where Claude Code is going to help. Remember this artifact that I created earlier? this portfolio style variations. I can just click into this artifact in the web version here and I can go back through all of those different styles that it gave me for my personal website. So, I can decide which one that I want to use and then I can take that over, hand it to Claude Code, and it can put the final polish on and deploy this website. So, I actually decided that I really like the studio hand website here. I don't like everything that it's added and I definitely want to make some modifications to it. But this is a great starting point for the styling. So what I can do in this project is I can say I want you to take the studio hand styling and turn it into a JSON object so that I can hand it off to my developer. And if I send that off, it's going to create an instruction manual essentially that Claude Code can understand extremely well. And it looks like it's doing it for the time zone slider. So, I'm actually just going to stop it and I'm going to say actually do it for the personal website in style 4. Okay. So, now it's picked up the studio hand and it's actually listing out everything that it needs to for this style. So, it's including things like the theme, all the colors for the different styling, the typography, everything that we need to pass this style off to my developer, which is Claude Code in this case. And it's including all of this generic information. I'm just actually going to stop it again and I'm going to tell it don't include generic information. Just include the exact styling, how we do the header variations, how we have all of the colors and everything like that, the layouts, spacing. That's the main thing that I want to focus on. Okay. And now it should be able to get it right. We had a lot in this chat. And also, we're on set 4.6, which isn't quite as good as Opus. So, now that I've been specific, it's just giving the style for studio hand, which is great. This is exactly what I wanted. Okay, so it wrapped up with this. Now I'm just going to copy this. And now in our Claude code section, I'm just going to tell it to create a new folder in here. So now I'm just going to tell Claude, I want you to create a folder within tutorial called personal website. So now I'll send that off. Now I could be creating this folder. I'm just showing you how I would typically go about this. Yes, this might seem lazy to you, but I honestly think that's a lot quicker what it just did there than me going in manually creating the folder and such. So perfect. It created that folder. Now I'll start a new session and I'll drill into that folder. So we got tutorial, we have our Claude MD in here telling Claude that this is basically our projects directory. And then if I go to personal website and hit open, now I can begin working on our personal website.

So we're going to go back to the GCAO prompting sequence. And this is going to ensure that I get all the best results with this out of the gate. So first I'll start with the goal. So for the goal, I just said I want to create a personal website using Astro and Tailwind. Astro is a framework for creating very simple websites and Tailwind is a styling framework. I want it to have a blog and a homepage. On the homepage, we should have a featured section. We're also going to be using Netlifi to deploy this. So, I want it to be easy to deploy and I want to stay on brand with these styling characteristics here. And I'm going to give it those styling characteristics that we generated from our artifact. Perfect. So, this is getting pretty long already, but let's go ahead and give it the C, which is context. And this is going to be where I give it some information more about me and what I want to have on that website. All right. So now I've given it some context for the website where I say for actual content and for context around this website. My name is Carter Sira. I've spent the past decade creating tutorial content on productivity software helping individuals get more productive in the workspace but also in their personal life. And I've amassed tens of millions of views on YouTube reviewing and teaching software. My main focus right now is helping people use AI tools. Now I've given it some context. I've just talked more about me and what my offer is. All right. Now I've given it some actions. So I've told it exactly what I want it to build, all the pages that I want. I also told it the main goal of the website and how I want it to store those blogs. And now I've just said we will deploy with Netlefi. I want to be able to post blogs when an API endpoint is hit. Meaning I can talk with this thing through Claude later down the line. And I can actually have it help me write the blogs and then Claude can post the blogs for me. I also want to be able to patch and edit those blogs. meaning I want to be able to update them. Essentially, we'll need to add an API key to the environment variables. And blogs can also be in a draft state. So, the API key is basically my password to be able to upload and edit blogs, and I'm going to be able to securely give that to Claude.

So, now I've made sure that I'm in my personal website directory, and I've written this monster of a prompt, and I'm going to send this off and see what Claude can create. And now, it's running me through a Q&A process so that it can get some more context. And I didn't include some things that I should have already included. So I'm going to go ahead and give it this information. So the link is aifoundations.io/apply. For the blog API endpoints, should the blog content be stored as a JSON file on disk? Simple. Works great with Netlefi or do you want a database like SQLite? I'll go with JSON files since that's what's recommended. And for the API endpoint that creates/edits blogs, what should trigger a rebuild on Netlefi after a blog is posted via the API? And this is getting a bit out of my wheelhouse here. So, I'm just going to ask it explain these options a bit more in depth. I want to know the pros and cons and what I should end up doing for this based on my use case. So, I'll go ahead and submit that. And I'm going to go with the recommended approach here. And now it's actually gone into plan mode and it's writing out a plan to create this website. And now Claude has drafted up this beautiful plan that I can scroll through. And I can even click on the sidebar here to go to different sections of the plan and read through it to make sure it's exactly what I want. And then when I get down to verification here, it's just going to tell me exactly how I can run this and test it. So I'm going to approve the plan. And now Claude Code creates a to-do list and begins scaffolding my project. All right. And just like that, it's completed my website. Now I can click preview in the upper right hand corner here and I can see what it built. I'm just going to drop down this debugging panel here and I'll scroll through the site and we'll see what we got. So, right now it's on the blog page. If I go to the homepage, though, as you can see, it got the exact style that we were looking for. So, I just said remove the section with the cards and the things that I can do and the about section entirely and make the top hero section a two column layout. So, right now it's this single column layout. I want this to be two columns. Maybe if I go wire screen, I don't know. I think that that looks kind of weird. So, I told it to make that two columns and then instead of being side by side. So, let's go ahead and send that off.

All right. So, the site is complete and I will go ahead and walk you guys through it. But, I ended up spending a couple hours getting this right and getting it to be exactly what I wanted. But overall, the site came out beautifully and I created a newsletter where people can sign up and people can also access the latest news in AI in the form of a blog and get notified when I come out with a new blog every Friday. Clicking into one of these blogs just opens up a nice widescreen view. And I also have ways to connect with me on the homepage, like watching my free YouTube tutorials, joining our AI Foundations community, or working with us directly with our done for you services. If you guys want to see the finished product, you can check it out at carters.com. But now I'm actually going to show you how to deploy this website. And to do this, we're going to use the Netlefi connector by clicking the more options and then going to connectors. And then next I'll hit add connector. And I'll just type in Netlefi. And then I'm going to click to add the Netlefi option. This is going to allow me to create, deploy, manage, and secure my websites on Netlefi. Now, if you already have a Netlefi account, it should open up and then you can authorize. But if you don't have a Netlefi account yet, you'll have to create one. Now, once I have Netlefi connected, it's as simple as going to the more options, hovering connectors, and making sure that Netlefi is turned on. And then I can simply say, I want you to deploy this website on Netlefi. And it may ask you for a few more commands. I'm just going to allow these. And just like that, the deploy is building. And my site is going to be live at this URL. In a moment here, I'll show you how to set up the environment variables, even though Claude did most of this for me already. And the nice thing about this website is these blogs that I've created here. I'm going to turn this into a skill that can then run in Claude to help me write these blogs. and I'll be running that version of Claude Code in a remote Claude Code terminal. So, you're going to get to see how I set that up as well. But now, let's check out our site. I'll just click this link and open link. And boom, the site is now live.

Now, to set up my domain name, I'm going to go into Netlefi and click on my Netfi project. And instead of this generic domain name, I want to use carters.com. So, I'll click on domain management. I'll hit add a domain. And then you can either buy a domain through Netlefi or you can add a domain that you already have. So I'll enter my domain and hit verify. And then you can just hit add domain. And it's going to say pending DNS verification. You can click on this to get instructions on how to connect it to your domain.

Next, let me explain the environment variables that are powering this website. So environment variables, as I touched on briefly before, are used as passwords to authenticate certain actions that you want the website to take on your behalf or that you want to request to the website. So now I've asked my cloud code session, tell me the keys I need to add. And it just says to add an API key. So this is whatever secret you want for the blog API. To do this on a Mac, you can just open up a terminal like so. And then you can type in uyuid genen one word and then hit enter. And that's going to generate a unique ID. Now I'm not going to use this exact ID because this needs to stay secret. So I'm not going to be sharing the exact ID that I use in this video, but just type in uyuid genen if you need an API key and then you can copy that key. Another way to visualize this is this is the key that is going to give Claude access to read and write these blog posts for me. I'll go into project configuration. And then if I scroll down a bit here, I should be able to find the environment variables tab. I can just click on that. And then here it says no environment variables set for this project. I'm just going to add an environment variable. And we're going to add a single environment variable. And then here I'm just going to type it in exactly as Claude said, API_key for the key. And then for the value, this is where we're going to paste in our UU ID gen that we sent off. So, I'll paste in the one that I generated. And I'm going to check this box that says contain secret value. And then if I scroll down here, you'll see that it added the value for production, deploy previews, branch deploys, preview server, and agent runners. And then we don't need it for the local development netifi CLI. I'm just going to hit create variable. So now the API key is live. The next ones that we need to add are the resend API key and the resend audience ID. And these can be generated through resend.

Now, since we just added these environment variables, I'm going to tell Claude to redeploy and use the environment variables that are already in the project. And now I've told Claude to redeploy Netlefi using the environment variables that I added in there so that it picks up those changes. And now it's going to run the deploy command with the Netlefi connector that we have installed. So I'll hit allow once. And now that it's done with that deploy, I'm going to test out an email to see if it actually receives it on resend. So, I'm going to accept and sign me up. And it says you're in welcome aboard. But let's check resend to make sure. And as you can see, our newsletter has one contact. And if I click into here, it did pick that up. And now, if I tried to refresh and then I type that email in again and try to send this off, it's going to block me because I've already signed up.

Now, I'm going to tell this session to create a skill that will help an agent actually post blogs, delete blogs, and manage my blogs for me so that I can set it up in a remote clawed virtual machine. And then I can just talk to this thing whenever I want to update the site. So, I just say, I want you to create a skill that uses the API key as an environment variable, but does it in a secure way. I'm going to be setting this up as an environment variable in a clawed virtual machine on anthropics virtual machine service. I want you to have the skill walk me through creating a draft of the blog first that I can visually look at for pulling full blogs if I have questions. But never post without asking the user for permission. Once the user gives permission, you can post. Also, make sure that you ask the user permission to delete or update blog posts. And also, we need some terminology in there for drafting and publishing as well. So, I'll go ahead and send this off and it should go ahead and create that skill for me. And then I can give Claude that API key in a secure way. And I'll show you how to do that. And now it's creating that skill. And as you can see, it references the API key using blog API key in shell commands. And I'm going to show you how to set that up in Claude's environment so that it can securely use that API key without actually reading it or compromising it. And now it says that the skill is ready.

Now I'm actually going to tell it to move this skill into this project's directory for personal website. So I just said move this skill into the projects directory for personal website so that when I use GitHub I can actually deploy this with the skill. Also let Claude know that it's going to see the entire codebase but it should just focus on the skill. If the user wants to write a blog, Claude will be connected to a GitHub repository for this skill and for this codebase. So I'm going to send this off. So, it's added that skill to the proper location. And now we have to get this code on GitHub. So, I'll make sure that I'm logged into GitHub. And I'll go to github.com/new. And then I'm going to create a repository name. And I'm going to choose the owner. And I'll just call this GitHub repository Carter Srirach. And then for the description, I'll just write my personal website. And this is really important here. We want to make sure that we choose the visibility of private so that this isn't public to everyone on GitHub. Now I'll create the repository and I'll just copy the repository URL right here. And then I'll jump back into cloud code and I'm just going to say create an initial commit to the branch called main. If main doesn't exist, create it and include all of the changes to this website that we have so far. Make sure to add a git ignore file to hide sensitive information. Make sure to also include that skill so that Claude can see it when we connect to this repository. And I'm just going to paste the repository URL and send this off. Even though we do have a private GitHub repository, we still want to go through the best practice of adding a.get ignore file. And usually Claude does this, but sometimes you have to tell it to do this. It says everything looks correct. The skill is included. Perfect. So now Claude will be able to know how to use the blog manager. And it says all done. Here's what was set up. 41 files were committed on the main branch. We pushed it and the skill is included. And when I go to that GitHub repository, I can actually see all of the code that is running my website. I can also see the Claude file, the skills, and the blog manager MD. So, this is perfect. This is going to be instructions for Claude to know exactly how to post blogs to our website.

Next, I'll copy that environment variable blog API_key, and I'm going to open up a new session in Cloud Code. And we'll click on the little local button down here. And then we're going to add a cloud environment. You could also just edit your session here on local and you could add that as an environment variable here and you could just put the API key, but I want to show you guys how to actually set this up with a cloud environment so that you can do this wherever you go when you're on the go. You don't have to have your computer on. So, I'm just going to add an environment. And here I'm going to paste in the API key. And then I'll add the API key right here after the equal sign. And now I'm just going to call this blogger. And I'm going to add this API key. And then I'll go ahead and create the environment. So I've just given it full network access. Now I'll save the changes. And now what that's going to do is it's going to allow me to select a repository. And I'm going to search for my website. And it's right here. So now I'm just going to tell it to try to read me the blog posts. Read me the blog post on our live site. And boom. Now it has given me the blog post. So I could continue working with this. Now, since I have this in a cloud environment, I can open this up on the go on my phone. So, I can open up cloud code on my phone and the code doesn't even need to live on the phone for me to be able to access this. I can just change my cloud environment to blogger. Then I can choose the proper GitHub repository. And I could tell it here, remove the blogs on my website, move all of them to draft, and then I could send that off. And because I've given it the GitHub repository and the right environment with the blogger, it's now able to do all of these tasks without me having to have the codebase live. So, it's listing the blogs and it said, "I moved the four publish posts to draft status."

Now, what if your API has an issue and you actually need to update the codebase? That's what I'm going to show you how to do next. I'm going to show you how you can deploy the changes automatically once you push them to GitHub. So, in order to do that, you're going to go into Netlefi. You're going to click deploys and then you're going to connect this up to your GitHub repository. I can just click on deploy settings. Then I'll click on continuous deployment and I'm going to link a repository. So we'll link to an existing repository. I'll click on GitHub and then I'll just type in my repository name here. If it doesn't show up, I can just hit configure Netlefi on GitHub. And then in a separate window, GitHub is going to open and I need to select the repositories that I want Netlefi to have access to. So, make sure you hit select repositories, add the right repository, and then hit save. Now, if I erase what I had in here, and I just try to type it in again, I should see that Carter Sra is now in here. So, I can click on it, and we'll keep branch to deploy on main. And I'll hit deploy Carter Srirach. And now it deployed the changes that the cloud virtual machine added, which it moved all of these to draft by default. So, it says coming soon. Stay tuned. And now all future changes can automatically update the website. For instance, from my phone, I could just tell it to change something like instead of every Friday, maybe I want this to be every Monday. I can just say update the header in the website to say every Monday instead of every Friday, then push that to the main branch on GitHub and it will autodeploy since we set that up. And as you can see here, right now it says every Friday. Let's go ahead and send this off and Claude will complete the change. And I just want to hit home. The beauty of this is is that you can be anywhere and you can update the code on your website because it's running on Anthropics server when they're actually doing these edits. It doesn't need to be running on your computer. It doesn't even need to be running on your phone, but you can communicate with your codebase directly from your phone and make changes to your website just by chatting with Claude. So, it's made that change. Now, it's committing and pushing to GitHub. So, it says it should auto deploy on Netlefi. Now, let's just go ahead and refresh and see if this changed on our website. And boom. As you can see, it changed it to every Monday. All done directly through my phone.

All right, now, my site has everything that I want. And I even added our first news. So, if I click into it, it even generated these nice images for the feature image, like this one of the butterfly for Project Glass Wing. And I've instructed it in the blog writing skill to do all of the research that it needs, identify topics, and then propose a topic to me. They even added these nice charts here and screenshots from the web.

Next, I'll show you how to schedule tasks in Clawed Code. So, if you open up the side panel here and then you go to scheduled while you're on the code tab, you can create a new task. And then you have to decide, do you want this on local like your actual computer here? This will require it to be on for it to run that scheduled task. or do you want to run it as a remote task in that environment that we set up earlier for the blogger? I'm going to go with remote task so that this can always run. I'm not relying on my computer being open or me actually sitting at my computer. Why take up this machine when I can just spin up a machine remote? So I'll click new remote task. And now I can give it a name. I'll call it daily blogs. So I've just said you are now in blogging mode. This is a scheduled task where you will run through the research flow independently. Find three topics that are timely and write drafted blog posts about them. Then send me an email using the Gmail connector when you're done. Never publish the blogs. Just send me a link in my email once they are ready. Use the Gmail connector to email me atcart.com from my Gmail account with a daily summary and links to the drafts. And then I'll just select my repository. And I'll make sure that I'm in the proper cloud environment. And we'll just say we want this to run daily at 6:00 a.m. so that I get a nice morning briefing from Claude when it's done. And then if I scroll down here, I see that we have some connectors. I'm just going to turn off some of the connectors that I don't need. We actually don't want Netlefi in this one. I just want it using the API. And then for Gmail, I guess that's the only connector that we want so that it can send us an email. And then I'll go into permissions here. And here I can allow unrestricted branch pushes, but I'm not going to do that. I don't need it pushing branches. I just simply need it doing the task of writing the emails through the API. We don't want it actually updating the codebase on a schedule. So now I'm going to create this and the next run will be tomorrow at 6 a.m. So this is just going to run like clockwork, constantly giving me drafted blogs every day that I can review and then I can edit alongside Claude later down the line.

Now if I click into here, I can also go ahead and run this right now and I'll show you what this process looks like. So let's just hit run now. And as you can see, we have a manual rundown here. If I click into it, it's actually running a session here. So, I can see exactly how it goes through this process. As you can see, it's reading the skill files that we've added, like the research.m MD that tells it how to do web research and the writing.md that passes along my style. And if I go show more here, we can also see some of the other things that it's using like the dither much tool, which allows it to generate those images and actually create that dithered style on them. blog manager kind of tells them how to post the blogs. And first it's testing to make sure that all of our keys are present and they are. And it's doing a tool search of Gmail here, making sure that it has access to those tools. And now it's created this to-do list right here. So the to-do list that it derived from my prompt here that it's going to run every day at 6 a.m. is number one, researching trending AI topics via Perplexity Sonar. That's what it's using here with the research.md. Then after that, it's going to select three distinct timely angles. Micro, mezo, and meta. Meaning, what is the most granular thing going on today? And maybe it's a product update. Maybe it's a new model. So, as you can see, this is what it would look like. Micro, the agent simplification wave, co-work, cursor 3, muse, spark, all within a week. Mezo, the super app consolidation era. And this is talking about chat GBT valuation. And then meta, the junior dev compression. This is more trend focused. It's going to talk about overall trends that it's seeing, not necessarily the day-to-day stuff. Now, it's drafting all three of these in parallel with the writing agents. And if I scroll through here, I can see as it goes through the to-do list. But normally, this could be happening totally in the background. And now it says create drafts via blog API status equals draft is complete. And it's going to email me a summary with draft links. And since this would normally be happening in the background, since this is a scheduled task, all that I'm going to see is those emails coming in. And then I can go in on a more focused session and I can talk to Claude later and I can iterate on those blogs using the blog API that I gave it access to. It's actually really good at updating the blogs. And the Project Glasswing blog that I added on my website already took a bit of editing and a bit of human oversight. But I would say that the AI gets it 70% of the way there. And just like that, the agent sent me the email with the summaries and the links to view the preview. So now let's go into the preview URL for this one right here, the Junior Dev Compression. And I have this password gated so that not just anyone can view my blog posts when they're in draft. I'm going to sign in. And then we have the entire blog post that it wrote out here. Let's just take a look through it. I would say it looks pretty damn good. Now, what I would do is I would read through this further. I would have it generate the feature image. It actually mentioned that it didn't generate the feature image for me in these drafts because it didn't want to burn through my API credits since it uses an image generation tool. If I wanted it to generate them anyways, I could just have it do that. But I don't mind going through during the publication process and just making sure that everything looks good. And I could do that process through my phone here. I could just say something like, I want you to find the junior dev compression blog draft and I want you to generate some featured image options for it that match the other blogs feature image style. And as I've said already, this could take place anywhere that you have your phone with you and you have an internet connection. You don't need to be on your computer. I'm simply screen sharing this with you so that you can see what my phone looks like, but this could be anywhere. So, think about the workflow. You could be at the beach. You could be on a road trip. You could pretty much be anywhere. Receive that email from Claude. It's just going to work like clockwork every day. And then you could go in on your phone, view the published drafts, and then you could text back and forth with Claude Code from your phone and have it put the final touches on. Once you've architected the system, it works for you in the background, and you can take it with you on the go.

All right, now it's come back to me with those preview images. So, I can simply click to open any of these links here, and I can view what they look like, and it's going to show me the images on my phone. I'm going to say let's go with option three and add it to the drafted blog on the website as the featured image. And it says it's done. So I can refresh my website over here. And as you can see, it's added that nice background image for me. Now, I could also read through this and I could ask it questions about how I can improve this or I can work with this in a more meta way and I can have it actually improve its processes by editing its skill files directly through the phone. So, the first few times around with the workflow, you might not have the exact results that you want, but after a while, Claude can hone the skill, chip away at it, and get the perfect process down. It's just like training a new employee. When someone comes into your company, they don't have all of the skills that they need, but you can actually craft these skills in a strict SOP. And the beautiful thing about Claude is it follows the SOP a lot better than a human would. This is a 14step workflow that it can go through if I wanted to actually write the blog, research everything, create the images, do the final touches, check for AI sounding terms, and humanize it. All of those things are done through a structured process. So any kink in the chain or any point that I want to change about this SOP or the skill for Claude, I can do that on the go.

Now the final method for interacting with Claude code is to use the terminal directly. And this is going to open up some more possibilities in terms of commands. It's a lot more flexible and familiar to a developer or someone who's already used to the terminal. And I found that this is the quickest and best way to interact with Claude if I have time to sit down at my computer and actually use it. So, I'm going to CD into my tutorial project. And then I'll list what's in here. I can see that I have personal website in here. So, I'll CD into personal website. And now I'll clear out my terminal. And I'm just going to type in Claude and hit enter. And now Claude code is active right here in my terminal. You can adjust the model by just typing in /model and hitting enter. And then you can arrow key through and you can decide which model that you want. I'm going to go with the default recommended. And then I can type in slasheffort and I can decide on the effort. When I hit enter, it's going to show me the effort level that I'm currently on. But if I want to change the effort level, I can just type in slasheffort and then I can type in exactly what I want. So let's say I wanted to go with low effort, I could just type low and hit enter. And then it's going to set the effort to low. Let's go ahead and change it back to high. And then when I hit slash again, as you can see, if I arrow down through here, it's going to show me all of the commands that I can use. And then if I want Claude to do something, I can just type it in right here. So I can just type out a simple prompt like assess my website and check for ways that we can enhance the speed and the SEO. And then I could hit enter. And inside of the terminal, we also get some extra commands like /btw, which stands for by the way. This can ask a side question that doesn't interrupt the flow so that I can get an update. So if I type in /yway and then space, I can just type out my question. How is it going so far? And then I can hit enter and it's going to answer that in parallel while this task is still running. This by the way session can see what's going on and it can tell me what we've done so far and what it's going to work on next. Then it'll answer the by the way question. And if I hit escape now it's actually answering my question. And it's giving me the biggest wins for speed improvements and SEO.

I hope you've enjoyed this beginner to expert guide using Claude. All the links are in the description below along with the chapter so that you can click back through this video and find exactly what you're needing more help with. If you made it all the way to the end, leave a comment below and let me know what your thoughts were on this training. I'm genuinely curious to hear how I can improve and what I did well so that I can make better claw tutorials for you in the future. Leave your questions and comments below. I'm looking forward to hearing from you. And subscribe if you don't want to miss the next video from me. I also want you to like this video if it helped you. If you've made it all the way through this far, you probably enjoyed it. So, go ahead and drop a like. I would really appreciate it.

Get Our Mobile App

FULL Claude Tutorial For Beginners in 2026! (FULL COURSE)

Transcription