Watch These to Master Claude

You have watched an AI video before. Maybe twelve of them. And you still could not tell someone what you actually learned.

That is not a you problem. Most AI content is made to impress you, not to teach you. You finish it feeling behind instead of capable, so you save three more videos and watch none of them.

So here is a small list instead of a big one. Five free YouTube videos, in the order to watch them. The first one assumes you know nothing. The last one has you building a real app on a phone. Nothing in between is skippable, and nothing costs money.

By the end of this page you will know exactly what each video teaches, why it comes where it comes, and what to build after you watch it so the information actually sticks.

First, why does this even matter?

Picture two people who both graduated last spring with the same degree and the same empty GitHub.

Person one spends the next three months applying. Four hundred applications, same resume, a little more discouraged every week.

Person two spends one weekend on this list. Then they build a small thing. Then a slightly bigger thing. By month three they have four projects they can demo live, they can talk about how AI tools actually work instead of repeating headlines, and every interview turns into a show and tell.

Same degree. Same starting point. The only difference is that person two stopped consuming and started shipping, and these five videos are the shortest path from one to the other.

You do not have to do this in one night. One video per sitting is a real pace. Just do them in order, because each one leans on the one before it.

Step 1: Learn what Claude actually does before you try to be clever with it

Video: FULL Claude Tutorial for Beginners in 2026 (Become a PRO) by AI Foundations

What this means in plain English: Claude is an AI assistant made by a company called Anthropic. You type something, it types back. That part you already know. What you probably do not know is the other 80 percent: Projects, file uploads, custom instructions, artifacts. This video walks all of it, start to finish, in under an hour.

Two words worth defining right now. A prompt is just the thing you type. Not a magic spell, not a secret formula, literally the message you send. And an artifact is when Claude opens a separate panel next to the chat and builds something live in it, like a document, a chart, or a working web page you can click.

Why it matters: Every other video on this list assumes you are comfortable talking to Claude. If you are still writing one line prompts and getting one line answers, the advanced stuff will feel like magic you cannot repeat. This is the base layer.

The beginner version of how to do it: Open claude.ai, make a free account, and watch this with the tab open next to the video. Pause and copy what they do. Do not take notes. Notes are how people avoid doing the thing.

Worked example: after watching, open a Project (a folder that remembers context across chats), upload your resume and three job descriptions you actually want, and ask Claude to tell you which specific skills those postings share that your resume does not mention. That is a fifteen minute exercise that gives you a real answer about your own job search, and you will remember how Projects work forever because you used them for something you cared about.

New here? Quick answer
Do I need the paid version to follow along?
No. The free tier is enough for this video and most of the next one. You will hit usage limits if you go hard for two hours straight, which just means waiting a bit. Upgrade later, once you know what you would be paying for.

Step 2: Turn Claude from a chat window into something that works on your actual computer

Video: Learn 80% of Claude Cowork in Under 20 Minutes by Jeff Su

What this means in plain English: Cowork is Anthropic's desktop app. The difference between it and the chat window is the difference between texting someone advice and handing them your laptop. Jeff walks through the seven core capabilities: local file access, persistent memory, connectors, skills, Cowork Projects, and scheduled tasks, with real examples you can try the same day.

A few of those need translating.

Local file access means it can open and edit the files sitting in your folders, not just files you upload one at a time.

Persistent memory means it remembers your preferences between sessions instead of starting from zero every conversation.

Connectors are links to accounts you already use, like Gmail or Google Drive, so it can read from them with your permission.

Skills are saved workflows. You do a multi step task once, then tell it to turn that into a reusable skill, and next time it is one sentence.

Scheduled tasks means it runs on a timer without you sitting there.

Why it matters: This is the first moment where AI stops being a thing you talk to and starts being a thing that does work while you are not looking. A daily 6am inbox task, for example, can read your email through connectors, apply the rules you saved in a local file, and draft replies using what it remembers about how you write. That is a different category of tool than a chatbot.

The beginner version of how to do it: Download the desktop app, point it at one messy folder on your computer, and give it one boring job. Boring is the point. Boring jobs are where you notice whether it actually worked.

Worked example: take the folder where every screenshot and PDF you downloaded this semester went to die. Ask Cowork to read every file in it, rename each one by what it actually contains, and sort them into subfolders by topic. Watch it happen in your Finder window. That single task is what makes the concept click, and now you have a clean folder.

New here? Quick answer
Is it safe to let an AI touch my real files?
It asks before it acts, and you approve each step at the start. Do what you would do with any new tool: point it at a copy of a folder first, not your only copy of your thesis. After a few runs you will know how much rope to give it.

Step 3: Go where the real leverage is, the command line

Video: CLAUDE CODE FULL COURSE 4 HOURS: Build and Sell by Nick Saraev / Maker School

What this means in plain English: Claude Code is Claude running inside your terminal, which is the plain text window on your computer where you type commands instead of clicking buttons. Every Mac and every Windows machine has one. It looks intimidating and it is genuinely not.

Why does the terminal version matter more than the pretty one? Because there, Claude can read your whole project, write files, run the code, see the error it caused, and fix it. That loop is the entire skill.

Why it matters: This is the one that changes what you are able to make. It is four hours, which is exactly why most people scroll past it, and exactly why the people who watch it pull ahead. Four hours is one Saturday morning.

The beginner version of how to do it: Do not watch it like a movie. Split your screen. Video on one side, terminal on the other. Build the thing while they build the thing. If you fall behind, pause. You will get through it slower than four hours and that is correct.

Worked example: build a tiny command line tool that takes a folder of your class notes and turns them into a quiz. Twenty questions, answers at the bottom. It is small enough to finish in one sitting, real enough to use during finals, and it teaches you the full loop of asking for code, running it, hitting an error, and having Claude fix the error.

New here? Quick answer
What if I break something on my computer?
Make a new empty folder and do all of this inside it. Claude Code works in whatever folder you start it in, so a fresh folder means the worst case is deleting that folder and starting over. Nothing outside it is at risk.

Step 4: Build an agent, which is the skill people are getting hired for right now

Video: How to Build an AI Agent with Claude Code by AI Master

What this means in plain English: An agent is AI that does a multi step job on its own instead of answering one question. You give it a goal, it figures out the steps, uses tools to carry them out, checks whether it worked, and tries again if it did not. A chatbot answers. An agent finishes.

Do not skip this one. Steps 1 through 3 teach you to use AI. This one teaches you to build with it, and that is a different line on a resume.

Why it matters: "Familiar with AI tools" is on every applicant's resume now, so it means nothing. "I built an agent that does X" means something, because the interviewer can ask you how and you will have an answer. This is the highest ratio of hireable skill to hours spent on the whole list.

The beginner version of how to do it: Pick a task you personally repeat every week. Weekly, not daily, so it is big enough to be worth automating and small enough to finish. Write down the steps you take, in order, in plain sentences. That list is your spec, which is just a plain description of what you want built. Hand it to Claude Code and build from there.

Worked example: an agent that runs every Sunday night, pulls new job postings for one specific title in one specific city, scores each one against your resume out of ten, throws out anything under a seven, and writes the survivors to a spreadsheet with a one line note on why it scored well. You now have a portfolio project, a talking point, and a job search that runs itself.

New here? Quick answer
Do I need to know how to code to build an agent?
You need to be able to read code well enough to tell when something looks wrong. You do not need to write it from memory. Your real job here is being specific about what you want and testing whether you got it. Being specific is the skill.

Step 5: Ship something people can hold

Video: How to Build Mobile Apps with Claude Code: Full Course by Nick Saraev / Maker School

What this means in plain English: Same Claude Code from step three, pointed at a phone app instead of a script. You will hear React Native and Expo. React Native is a way to write one app that runs on both iPhone and Android. Expo is the tool that handles the annoying setup and lets you see your app on your actual phone while you build it.

Why it matters: A website is impressive. An app someone can open on their phone is memorable. When you hand an interviewer your phone with a working app on it, the conversation stops being about whether you can do the job.

The beginner version of how to do it: Build for one person with one problem, and make that person someone you can text. Scope it down until it is almost embarrassing, then finish it. A finished tiny app beats an unfinished ambitious one every single time.

Worked example: a soccer pickup app for the group chat you are already in. One screen showing this week's game, a tap to say you are in, a counter showing how many players have committed, and a red banner when it drops below eight. That is four screens of work and about a hundred people would use it if you built it for your city.

New here? Quick answer
Do I have to pay Apple to put this on my phone?
Not to build it or test it. Expo lets you run your app on your own phone for free while you develop. You only pay Apple's developer fee if you want it listed publicly in the App Store, and that decision can wait until you have something worth listing.

Putting it all together

1. Watch the Claude fundamentals video and use Projects on something from your real life the same day.
2. Watch the Cowork video and give it one boring folder to clean up so the capabilities become real.
3. Watch the four hour Claude Code course with your terminal open beside it and build along, not after.
4. Watch the agent video and automate one weekly task you already do by hand.
5. Watch the mobile app course and ship one tiny app for one real person.

Five videos. Five things built. Roughly one weekend plus a few evenings.

The part nobody tells you

Almost nobody who reads this page will watch all five. Most will save it, feel briefly productive, and never come back.

That is the whole opportunity. The bar is not talent and it is not your GPA. The bar is finishing a list. Everyone has access to the same free videos, which means the only thing separating you from the person who gets the offer is whether you actually sat down and did them in order.

So pick a start time. Not "this week." A time. Open video one, put your phone in another room, and build the small ugly thing.

You are not behind. You just have not been handed the list until now.

If this helped, follow along at @compskyy for the rest of it. New builds, new tools, and the stuff nobody teaches you in the CS curriculum.