Claude AI Tutorial 2026: How to Use Claude AI for Beginners

Most people use Claude the same way. Open it, type a question, get an answer, close the tab. That's like hiring a superstar employee and only asking them to fetch coffee. But Claude isn't just a chatbot. It's a full system. One who can know who you are, write in your voice, plug into your tool, design your assets and run tasks on its own while you're doing something else. And in this video, I'm going to show you how to build that system from scratch. If you've tried Claude and thought it's fine but it just gives me the same generic stuff, that's a skill issue, not a Claude problem. In this video, I'll show you how to get the most out of Claude, how to use the new Claude design to create stunning on-brand design assets and how to even stop burning through your token limits. Everything in this video, prompts, templates, skill files is free inside our school community. The link is in the description. Now, let's get into it. Picking the right model and turning on web search. First thing, the model selector. You got three options. Haiku, Sonnet and Opus. Haiku is fast and cheap. Sonnet is your daily driver. Sharp, quick, handles 90% of what you throw at it. Opus is for when the answer matters more than the speed. Strategy, research, complex reasoning, that's when you switch to Opus. Be sure to match the model to the task so you don't hit your limits too quickly. For most things in this video, Sonnet is all you need. Now, the one setting most people never touch, web search. Go to settings and turn it on. Most people who complain about Claude giving outdated answers have never turned this on. Turn it on, leave it on. That's it. 30 seconds of setup that immediately makes Claude more useful. How to prompt Claude better. Here's the biggest mistake people make with Claude. They treat it like a search engine. Short question, expect a great answer. That's not how this works. When I typed, give me ideas to grow my YouTube channel, this is what I got. Generic, surface level, the kind of advice you could have just googled. Now, watch this. Night and day. Three things make the difference. What role you want Claude to play, the context of your specific situation, and the format you want the answer in. Here's a trick. Instead of guessing what context to include, just ask Claude what it needs. Add a line at the end. What do you need to know from me to do this well? It will ask you the right questions and the output will be better for it. Memory. Here's a feature most people completely overlook, and it's probably the highest leveraged thing you can set up in your entire Claude account. Claude builds memory from your conversations. Over time, it learns who you are, what you're working on, and how you'd like to communicate. So instead of re-explaining yourself every single chat, Claude already knows. You can see what it's storing, edit it, and even add to it manually. Go to settings, capabilities, and turn memory on. Now, here's the move that saves you months of buildup. If you've been using ChatGPT or Gemini, you don't have to start from scratch. Just click start import from a prompt you can paste into your old tool. It spits out a summary of everything it knows about you, and you can just import that straight into Claude. Set this up once. From that point on, Claude knows you. Artifacts. So far, everything we've done lives in the chat. Artifacts are different. Claude stops replying and starts building. Tools, calculators, dashboards, landing pages, interactive apps, all rendered live inside Claude. No code knowledge needed. Here's an example. I typed, Build me a content calendar planner for a solo creator. Inputs, how many videos per week, my main platform, and three content pillars. Output, a weekly grid that auto-assigns each pillar across the week. With a field to mark what's filmed, edited, and posted. Clean, simple, mobile-friendly. 90 seconds later, working planner. Then I hit publish and copied the link. Anyone I send it to can open the planner right now in Claude and start using it. Great to share with your team or drop in your community. Same move works with files. Drop in a PDF, a CSV, a screenshot, ask Claude to analyze it, and it builds you a full visual dashboard right inside the chat. Artifacts turn Claude from an assistant into a builder. Now onto one of my favorite features, projects. See, here's the problem with regular chats. Every new conversation Claude starts from zero. No context, no memory of what you're working on. Projects fix that. A project is a permanent workspace with its own files, its own instructions, and its own memory. Every conversation inside that project has access to all of it, every single time. Here's how I set one up for YouTube scripting. I upload three files, my brand guidelines, a sample script in my voice, and a one-pager on my audience. Then in the instructions, I write, You're helping me write YouTube scripts. My audience is small business owners learning AI tools. Tone is direct and conversational. No corporate language. Always reference the upload voice sample before writing. Now, when I open the project and type one sentence, draft a video about why most people get bad results from AI tools. Claude already knows my niche, my tone, my audience. The output is something I'd actually film. One thing worth knowing, memory is scoped. What Claude learns inside a project stays inside that project. And what it learns in regular chats stays separate. So you're not getting bleed between different workspaces, which is actually exactly what you want. Plus, projects can help extend your limits due to better caching. Skills projects give Claude the right environment. Skills give it the right process. Right now, every reoccurring task you have, writing emails, drafting posts, creating outlines, you're either retyping a long prompt every time or copying one from a doc somewhere. That adds up fast. Think of a skill as a saved playbook. You tell Claude once how you want something done, the tone, the format, the rules, and every time you trigger it, it follows that playbook automatically. No re-explaining. To add one, go to settings, search for skills, then hit the plus button. From there, you can browse Anthropic's official skills or you can upload a custom skill as a .md file, which you can usually find online. You can also just have Claude build one for you from scratch. Say something like, I want a skill that writes LinkedIn posts, tone is direct and punchy, short sentences, no corporate language, no hashtag spam. Every post opens with a one-line curiosity hook, delivers one specific insight, ends with a question or CTA under 150 words. Claude generates the skill file, I install it. Now, in any new chat, I just say, write me a LinkedIn post about X and it fires exactly like that every time. And when it doesn't work, just ask for it to be used. Connectors. Skills tell Claude how to work. Connectors give it somewhere to work from. This is where Claude plugs directly into the tools your work actually lives in. Google Drive, Gmail, Google Calendar, Notion, Slack. Click the plus icon, hit add connectors, and you'll see the full list. Most are one click. Once connected, Claude stops needing you to upload anything. It reaches into your drive, finds the docs, reads it and applies it on demand every time. Here's what that looks like in practice. Inside my YouTube scripting project, I connected Google Drive. Now, when I start a new script, I just say, pull my channel positioning doc and use it to inform this script. And it does. No uploading, no copy pasting, and no context loss. Connect the tools you actually use. Start with one. You'll immediately feel the difference. Research mode. You've got web search turned on from step one. But research mode is a completely different thing. Web search grabs a quick answer. Research mode is autonomous. You give it a question. It builds a research plan, runs multiple searches that build on each other, cross-references the sources, and comes back with a fully cited report. Takes 5 to 15 minutes, but the output is something you'd actually use. To turn it on, hit the plus icon in the chat bar and select research. Here's how I actually use it. I'll type something like, research the top five AI tutorial channels. And it comes back with citations, patterns, and gaps I can actually exploit. That used to take me half a day of manual research. Now it's done while I make lunch. Use it for anything that needs depth, overspeed, competitive analysis, market research, big decisions. Not for quick questions. That's what web search is for. Claude design. Now this will make your jaw drop. Claude is now a really powerful design tool. It's a relatively new feature called Claude design. And it takes a single prompt and turns it into a fully designed asset. Slide decks, prototypes, dashboards, wireframes, even video animations. All on brand, all editable, all inside Claude. Just click the design tab. And the first thing you'll notice is the template selector. Slides, documents, prototypes, wireframes. Pick your starting point. I'll show you slides because this is where it immediately replaces a tool you're probably already paying for. I typed, build me a five slide deck about what's new in Claude. Clean modern style, dark background. It asks a few questions and 30 seconds later, a full presentation. I can present it live directly from Claude, export it as a PowerPoint file or hand it to the client as is. And there's even a full editor. Click any element and you can retype it, resize it, recolor it exactly like you would in Canva or Figma. You can even ask Claude to change a specific element. But to make this design even more on brand, use design systems. You set this up once. Your fonts, your colors, your brand rules, and every asset Claude generates automatically follows it. Slides, dashboards, animations, all consistent. You can even build your design systems inside Claude Code and sync it directly to Claude Design with one command. Claude Code is a more technical, more capable version of Claude. And we'll cover how to use it soon. So subscribe to stay tuned. And through connectors, you can wire in image and video generation tools directly. So if you need a custom visual inside your deck, Claude generates it without you ever leaving the tab. Isn't that cool? What if Claude could do your work while you're at the gym, on holiday, or even asleep? That's Cowork. Add plugins and you turn it into a specialized expert for whatever you're building. Let's start with Cowork. Cowork is Claude with access to your actual computer, your files, your apps, your browser, not just answering questions, actually doing things. To set it up, download the Claude desktop app. Sign in and click the Cowork tab. First, create a dedicated folder on your computer. Call it Cowork Workspace. And point Cowork at it. Everything it builds, reads, and saves lives there. Then drop an aboutme.md file inside. Two short paragraphs on who you are and what you do. Here's why this matters. By default, Cowork doesn't remember anything between sessions. That file is how you give it continuity. It reads it at the start of every session. So you never re-explain yourself. Next, go to settings, Cowork tab, and set your global instructions. Things like always confirm before deleting a file. Keep responses concise. Think of it as the brief you'd give a new hire on day one. Now, connect your tools. Gmail, Drive, Notion, whatever you work in. Cowork doesn't just read from these. It uses them the same way you would. That's the setup. Once it's in place, you can point Cowork at almost anything. Organizing a messy folder. Pulling data out of a stack of PDFs into a spreadsheet. Turning scattered notes into a clean document. The sweet spot is work that's too big for a normal chat window and has multiple steps you don't want to babysit. Here's the feature that turns it from a tool into an employee. Scheduling. Inside any task, type dash schedule. Set the time and frequency and Cowork runs it automatically. The classic use is a Monday morning briefing. You set up one task, pull from your connected Gmail, Slack, and calendar. Figure out what's coming up this week and drop a clean summary in your folder. Set it once. Every Monday, you open your laptop to a full week of briefing ahead. Already written and you did nothing. You can schedule almost anything on a loop. Weekly research reports. Reoccurring file cleanups. A stand-up draft of whatever you can produce on a regular cadence. Set the cadence. Walk away. Come back to finished work. One thing to know. The desktop app has to stay open while Cowork runs and longer multi-step tasks eat more of your usage than a normal chat. So point it at the jobs that actually need it. Last thing. Dispatch. In the Cowork sidebar, there's a QR code. Scan it with your phone and now you can send tasks from anywhere. Out for a walk. Idea hits. Open dispatch. Type it in. Cowork queues it for the next run. Now let's talk plugins. A plugin is a bundle. Skills, connectors and slash commands packaged into one install. Hit the plus icon. Add plugins and you'll see everything available. You can install plugins like the Figma plugin, which makes Claude a Figma expert. Or Exa for improved web search capabilities. Stack these together and you're not just using Claude, you're running a system. And that's the full system. Most people will watch this, try one or two things and go back to using Claude like his search engine. Don't be that person. The whole point of this video is to make these features compound. Memory feeds into projects. Projects feed into skills. Skills feed into Cowork. And suddenly you've got something that runs without you. And if you want to go deeper on Cowork specifically, how to build a fully automated content system that researches, writes and posts to every platform without you opening a single app, that video is up next. Go watch it. I'll see you there.