Big Idea

The argument in one line.

Most people use Claude as a smarter search box, but its real leverage comes from the structural layers — projects, memory, integrations, and skills — that let it act on your behalf without constant re-explanation.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You have opened Claude at least once but still type fresh context every time you start a chat.
You are a solopreneur, freelancer, or small business owner who wants Claude handling real recurring tasks, not just answering questions.
You want a single video that covers every major Claude feature in the current desktop app, with no assumed technical background.
You have heard about Projects or Skills but never set one up.

SKIP IF…

You already use Projects, Memory, Connectors, and Skills daily — this is entry-level coverage.
You are looking for API usage, prompt engineering theory, or agent orchestration; the video stays firmly in the consumer UI.

TL;DR

The full version, fast.

Claude is not a search box — it is a coworker you hand real tasks to. The five levels are: (1) writing prompts that ask Claude to probe you before it writes, (2) setting up Projects with instructions and files so you never re-explain yourself, (3) enabling Research mode and connecting your apps via OAuth or Zapier MCP, (4) using Artifacts to build and publish interactive tools in plain English, and (5) creating Skills (markdown recipes) and installing Plugins (bundled skill and connector playbooks). The single highest-leverage first step: create one Project tonight with a page about yourself.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:54

01 · Cold open — the reframe

Hook: Claude is a coworker, not a search box. Promise of a complete beginner guide.

00:54 – 02:48

02 · Which plan do you need

Free, Pro ($20), Max ($100/$200), Team/Enterprise. Recommendation: start on Pro, upgrade only if hitting limits.

02:48 – 04:33

03 · The lay of the land

Chat vs Cowork vs Claude Code. Projects, Artifacts, Customize — a tour of the sidebar.

04:33 – 06:37

04 · Level 1 — Talking to it well

Prompt formula: who you are + task + rules + ask five questions first. Wedding toast demo.

06:37 – 08:12

05 · Level 2 — Projects

Job Search project demo: instructions + resume upload. Every chat in the project inherits context.

08:12 – 09:33

06 · Memory across chats

Memory settings panel, how to read/edit/pause. New: import memory from another AI provider.

09:33 – 10:47

07 · Level 3 — Research mode

Multi-step web research demo: best robot vacuums under $400, 19 sources, full cited report.

10:47 – 14:59

08 · Connecting your apps

Connectors directory: hundreds of apps. Zapier MCP for anything else. Canva OAuth demo. Calendar summary demo.

14:59 – 16:31

09 · Level 4 — Artifacts

What artifacts are. Mortgage calculator demo (interactive slider). Artifacts tab stores everything created.

16:31 – 17:47

10 · Building and publishing a quiz app

Quiz app built in one sentence, tested live (Roman Empire), published to a shareable URL.

17:47 – 20:41

11 · Level 5 — Skills

Skills = markdown recipes. Notes cleanup skill created live. The Skill Creator skill is built into Claude.

20:41 – 22:31

12 · Plugins — bundled playbooks

Plugin directory: Small Business, Sales, etc. Skills free; Plugins require Pro. Install demo.

22:31 – 24:36

13 · Cowork and Claude Code explained

Cowork = no-code work on local files, schedulable, mobile-kickable. Claude Code = most powerful, recommended even for non-developers.

24:36 – 26:59

14 · The models — Opus, Sonnet, Haiku

Opus 4.8 = flagship (default on Pro). Sonnet = faster, lighter on tokens. Haiku = speed-only. Effort slider: Low to UltraCode.

26:59 – 27:40

15 · Closing — the one thing to take

Set up one project tonight with a page about yourself. Everything else works better once you have done that.

Atomic Insights

Lines worth screenshotting.

Most people use 5% of what Claude can do — the gap is structural setup, not smarter prompting.
Ending any prompt with 'ask me five questions before you write anything' eliminates generic outputs and pulls your actual context out of your head.
Projects and Memory are separate stores on purpose — your job-search project context never bleeds into your personal chats.
Research mode runs a full multi-step web investigation and cites every source — it replaced two hours of manual research in the demo.
Claude can publish an interactive web app to a shareable URL in one sentence — no deployment step required.
A Skill is just a markdown file in YAML format — there is no code to write, and Claude can generate the file for you by describing the task.
Plugins are paid (Pro required), but building your own Skills is free on any plan.
Dropping to Sonnet for simple back-and-forth tasks is the easiest way to stretch your Pro token limits without sacrificing quality on harder work.
The effort slider controls how hard Claude thinks before answering — Max effort burns tokens proportionally faster.
Claude Code is recommended over Claude Cowork even for non-developers — it has more features and the floor for what you can build has dropped to near zero.

Takeaway

Five layers that turn Claude from a chatbot into a coworker.

WHAT TO LEARN

The people who get the most out of Claude are not better at prompting — they have set up the structural layers that make re-explaining yourself unnecessary.

Ending any prompt with 'ask me five questions before you write anything' is the fastest way to get personalized output — it forces Claude to gather your context instead of filling gaps with generic averages.
A Project is the highest-leverage setup step available: one page about yourself, dropped into a dedicated folder, means Claude carries your background into every chat you start there.
Projects and Memory are intentionally separate — your professional project context never bleeds into personal chats, and vice versa.
Research mode is not the same as web search — it runs a multi-step investigation across many sources and returns a cited report that would take a person two hours to compile manually.
Connecting Claude to the apps you already use via the Connectors panel is where it shifts from answering questions to taking action inside your actual workflow.
An Artifact is a finished, interactive object — a calculator, a web app, a published quiz — that you keep shaping through plain-language conversation.
A Skill is a reusable process stored as a markdown file: describe a task you do repeatedly, Claude generates the recipe, and you invoke it with a slash command from then on.
Claude Code is worth using even with zero technical background — the floor for what you can build has dropped enough that non-developers use it for content scripts, newsletters, and outreach campaigns.

Glossary

Terms worth knowing.

Project: A persistent folder inside Claude where every chat shares the same instructions, uploaded files, and context — so you never re-explain your background or goals.
Memory: Claude's automatic cross-chat profile of who you are and what you care about, built incrementally from normal conversations and separate from Project memory.
Research mode: A Claude feature that runs a multi-step web investigation — many sequential searches following threads — and returns a cited report rather than a single quick answer.
Connector: An OAuth or MCP integration that links Claude to a third-party app so it can read and act inside those tools from a chat.
Artifact: A finished output Claude creates in its own panel — a document, chart, working web app, or interactive calculator — that you keep editing by talking to it.
Skill: A reusable task recipe stored as a markdown file in the Customize section, invokable with a slash command, so you never have to explain a recurring process again.
Plugin: A bundled playbook that combines a set of Skills with the app connectors they need, installable in one click from the Plugin directory (requires Pro).
Effort slider: A per-model setting (Low / High / Max / UltraCode) that controls how deeply Claude reasons before answering — higher settings consume tokens faster.
MCP: Model Context Protocol — the open standard Claude uses to connect to custom or third-party integrations not yet in the native Connectors directory.

Resources

Things they pointed at.

01:20linktheaiaccelerators.com ↗

01:20productreprisesai.com ↗

01:20linkSkool — Systems to Scale community ↗

01:20linkAI Core newsletter ↗

13:00toolZapier MCP ↗

13:50toolCanva connector ↗

Quotables

Lines you could clip.

00:50

“Once that actually clicks, you stop asking it little questions, you start handing it actual jobs.”

Clean pivot moment — defines the video's thesis in one sentence→ TikTok hook↗ Tweet quote

17:37

“You just built and shipped a little AI application with just about one sentence.”

Punchline to the quiz-app demo — high concrete payoff→ IG reel cold open↗ Tweet quote

05:40

“Before you write anything, ask me five questions that would make this more personal.”

Immediately actionable — one sentence viewers can steal and use today→ newsletter pull-quote↗ Tweet quote

26:59

“Set up one project tonight. Drop in a quick page about yourself. Everything else that we just went through, it works better once you've actually done that.”

Strong CTA close — specific, low-friction next action→ TikTok hook↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

00:00I run most of my actual work through CloudNOW, and I don't just mean as a chatbot that I ask questions to. I mean that it remembers my projects, it builds real things for me, and it goes off, and it does work while I am doing something else completely different. And most people, honestly, are still using maybe 5% of what it can do.

00:15And over the last couple of months, this application has quietly turned into something completely different and nobody has handed out a brand new manual for this. This is the complete beginner's guide to Claude. So everything that actually matters, set up in the right way, fully up to date as of today.

00:29This is basically the video that I wish somebody handed me when I first started. So even if you've never opened Claude before, that's totally fine. You can just follow every single step with me today, and I know that it can feel like a lot right now, so I'm going to be taking it one piece at a time.

00:42By end of this, you'll know exactly what Claude is, what all this new stuff actually does, and how to use it without overthinking it. By the way, I'll have all resources and guides available for completely free inside of our Skull community. Link will be down below in the description.

00:54And before we touch anything at all, let me just reframe what this thing even is, because this one shift, it changes how you actually use it. Now most people, they just open up Claude, they type a simple question, they read the answer, and they close the tab. So they're treating it like just a smarter search box.

01:10And if that's all you do, okay, fine, it's good at that. But you're leaving the best part on the table. And the way that I like to think about this is it's less like a chatbot and more like an actual coworker.

01:22So something you hand a real task to. So assigning somebody to write this, research that, build me this thing, go through these files, and once that actually clicks, you stop asking it little questions, you start handing it actual jobs. Now some quick context on why I actually care about getting this right.

01:37If you're new here, I run three AI businesses, and I've spent the last couple of years helping hundreds of people build with it. So everything I'm about to be showing you is how I use it day to day, not theory. Alright.

01:48So let's get into the first real decision. Which version do you actually need? Because there's a few different tiers, and I'll keep this very simple.

01:54There's a free plan. This obviously costs nothing. It works on the web, it works on your phone and the desktop application, and it's just a great place to start learning.

02:03Now with this, even get the memory now, where it remembers you across chats, which just used to be locked behind paying. However, the catch is you will hit your usage limits pretty quickly, and the heavier stuff, it's going to be turned off, you're not getting the best usage out of it. And then there's going to be pro.

02:16This is $20 a month, and this is the one that I would be pointing almost everybody to. So you get a lot more usage, you get every model, and you unlock the heavier features. So you get the research, you get the plug ins, and connecting up more of your applications, pretty much all of it.

02:29So if you're getting serious, I would be starting here. And above that, there's going to be at a $100 a month, and even a bigger tier at $200 a month, which is what I use for my team. So that's just for people who live inside of Claude, use it all day, that sort of stuff.

02:41So this is gonna be for heavy research and heavy building. I will say that most people do not need the second tier of max, I hardly need it myself. But I would just recommend starting on Pro and only jump up if you're actually hitting your usage limits.

02:53And then of course, there are team and enterprise plans for companies, but that is a whole separate conversation. So for you, right now, it is free to learn, and Pro is available when you're actually just ready to put it to work. Now let me give you the lay of the land, because once you see how this is organized, everything's gonna be very easy afterwards.

03:09Now inside of Claude's desktop application, which we are using right now, obviously on the left hand sidebar we have the new chat, we have projects, and projects are effectively just going to be folders for all of your work. Artifacts, these are going to be, you know, where Claude can actually build things for you. Customize section, this is where you teach Claude any new tricks, and we will be hitting every single one of these.

03:31And up here at the top is the part that throws most people off. We have the chat, we have co work, and we have Claude code.

03:37Now it's the same Claude underneath, it's just three different ways to actually be working with Claude. So chat, as you would expect, this is solely going to be for just conversations. This is where you will likely be spending most of today.

03:51Cowork, this is where you can really be doing hands on work. Then the last one, we have code. Now code is going to be mostly for building software.

03:59I will say that I use it for about 99% of everything that I do nowadays. So whether it's creating scripts for any of my content, or creating newsletters to be sending out, or creating posts for inside of my community, or maybe just writing back to some incoming emails, or creating a plan that I need for maybe a new outreach campaign or a new ads campaign.

04:19Whatever it may be, I use Cloud Code for about 99% of everything I do now. Navigating back to the chat section, we can see that we are using Opus 4.8. That's the model doing the thinking, and we'll come back to that at the very end, and I'll tell you exactly when to change it, what you should be using, why.

04:32But for now, just leave it alone. Alright. So starting with level one, just talking to it, but actually doing it well.

04:38Because this is what separates people who think Claude is just amazing from the people who think that it is just okay. So it's the same tool, but totally different inputs.

04:46Now here's the trap with this. Most people, they type something like just write me a wedding toast, and then they wonder why it sounds like a greeting card. It sounds very robotic.

04:55And the model, it is not the problem. It just doesn't know anything about you, so it hands you the average of the whole Internet. So let me show you the way that I actually write a prompt, and I think of it really just in a few quick parts.

05:07So who I am, what I want, any rules, and then one little trick at the end that does most of the work. Here's my prompt. I'm the best man at my younger brother's wedding next month, and I need to write a toast.

05:16So write a toast so I can read out loud in about three minutes. It should be warm and a little funny, not cheesy, no cliches about finding your other half. Keep it under 400 words, and before you write anything, ask me five questions that would make this more personal.

05:28So look what I did there. I told it who I am and the situation. I told it the actual task, and just know you don't actually have to write out tasks and rules.

05:36Claude is smart enough to figure out it is a rule without saying that it's a rule. But anyways, I gave it the length, I gave it the tone, I gave it what's to be avoiding, and then the most important line, last but not least, is going to be right down here. So I told it to ask me five questions before it writes a single word.

05:53So what it's of course going to be doing is asking me those five different questions. So these are going to allow it to get better answers, a better ultimate output, because it's going to get more specific and about what I actually need.

06:05It's gonna be asking like any stories, because it does need context. So just having it probe and help me give it that context. And the main idea of why we actually do this is because we want to pull the good things out of our head.

06:15Again, as I mentioned, the context, so of course, we can be getting back 10 times better answers. Now I do this for almost everything and anything that matters, just end your prompt with ask me questions first to make sure we're gonna get a better answer. So inside of here, I'm going to be providing you with just a long list of fake information about Fake Brother that I do not have, and just like that, we're going to be getting our answer right back to us.

06:37Level two, making Claude actually know you so you stop re explaining yourself every single time. Now there's gonna be two things to do for this. It's gonna be the projects, and it's going to be the memory.

06:47So let me start with the projects because this is the one that I lean on I would say the most. So I mentioned this earlier, a project is more or less a folder, and everything inside it shares the same context. So let me just quickly make one.

06:59I'm gonna be calling this job search. Once you create that, you'll see we have the memory and we have the instructions. So we're going to first add instructions here.

07:06What I'll paste in is this project as my job search, use everything in here to help me tailor my resume, write cover letters, and prep for interviews. Here's some information about me, and here's how I want you to be helping me. Overall, I'm just saying to be very direct and honest and brutal at times.

07:21So we'll save those instructions. Now when you make one yourself, you need to be giving it a brief description, so what it's for and how you want Claude to behave in it. So that instruction right there, it's just going to be the description.

07:31So what this is for, how you want Claude to behave in this instance, and that instruction is now going to be sitting behind every chat that I open up and start here. Beyond that, you can upload files directly into this project.

07:43So I'm going to be dropping in one document about me, as you can see right here. So this is my resume, the kind of roles that I want, what I'm good at, what I'm not good at, and then just one page for this, it's gonna be plenty. So let me download this.

07:53So I've just uploaded this directly into Claude, and every chat I start inside of this project, it is already going to know all of that. So I can open one chat to rewrite a cover letter, I can open another one to just practice any interview questions, I never have to paste my background in here again. And you can just think of it like giving Claude a dedicated desk for just one part of your life.

08:12Next up is going to be the memory, which is the other half of this. So people, they often mix these two up. So let me be really clear about this.

08:19Memory is just Claude remembering things about you across the normal chats all automatically. So it quietly is going to be building up a little summary of who you are, what you care about, and it pulls that into new conversations, so you're not starting from zero every single time. So if you go into your settings, and we're going to navigate into our capabilities, you'll see we have our complete memory section right here.

08:40So you can read exactly what it has remembered. You can edit all of it if you would like to as well, if you ever need to. And then you can tell it to forget something or just pause it completely.

08:50So if we open up ours right here, can see it was updated about seventeen hours ago. It has quite a lengthy list of what we are, what we do, our companies, where we're based out of, and some other important context. Something that they just recently added is an option to now bring your memory from another assistant, from another AI provider.

09:06So if you use ChatGPT or anything else and it has a bunch of different memory on you, you can import it from here just by clicking on start input. So you're not gonna be losing any of that. You just bring it straight over into Claude, and you just, you know, seamlessly start using a different AI provider.

09:20Now one quick thing so you're not gonna be confused later, projects, they keep their own separate memory from your actual regular chats on purpose, so your job search stuff, it's not going to be bleeding into your personal chats. Just keep that in your back pocket and know that. Level three is giving Claude reach.

09:35So this is being able to give Claude the access and ability to go into the Internet and go on the web, and search for things, and then integrate with the applications that you already use. So using your text decks, so Slack, or Notion, or Gmail, OneDrive, whatever.

09:49First, we'll start off with research. Now, Plugged, it can already search the web on a very normal question, but there's a proper research mode that you should be using. So instead of just doing one quick search, it is going to go and run a whole investigation for you, search after search, following all of the threads, and then it can come back to you with a real report with its sources cited, so you can actually check them for yourself.

10:12So how you actually turn this on, you just click on this little plus button, and you could click on research instead of just the web search. Now we have both enabled. So let's try this out.

10:20I'm gonna say research the best robot vacuums under $400 right now. Give me a short comparison of the top three, the real trade offs, and link your sources. And now it's off working.

10:28In this part, it will take a few minutes. I'll just give it access to all of my different connectors. As you can see right there, it's just listing out every single one that I have given it access to.

10:38But it's just going to be reading a whole stack of sources and not just grabbing the first result. So it will run-in the background while you can do other things, and then it comes back with the entire entire write up. Additionally, you could click on this little section right here, and you can see exactly what it's doing right now.

10:51So it's researching the plan created. It's then going to gather all the 19 sources, and count them up, and go through each individual one. As it's furthering along in its research, we can click on any of these sources.

11:02It is going extremely in-depth. So we have eight results here, three results here for the best budget robot vacuums, 10 results, and it looks like it just gave us our output back.

11:12I mean, look at all these different sources. I probably would have spent two hours going through all of this stuff myself, or if I assigned somebody to do it, it would take about two hours minimum. Either way, we can click on this output document here, and here is our full report.

11:26So the best robot vacuums under $400, the top three compared. So there's the TLDR, the key findings, the details of each individual one with the cited sources, the quick comparison, you get the idea.

11:38Now moving on to the second half of Claude's reach, you can connect Claude's to all of the applications you already use, and this is the part that most people do not touch. So let me head into the customize section, and then we'll head into our connectors, and we can look at what is actually inside of here.

11:53So we can have and see my available ones that I have connected right here. Let's just go to the browse connectors and see what they have. So there's hundreds of applications in here sorted into categories, so you can just find your own.

12:03You can search up what you need inside of here, and we have different things like Slack and Notion, Asana, Linear. We have HubSpot, Stripe, Figma, Canva, Airtable, you name it.

12:15And with this, they are adding new applications almost every single day on top of this. Now some of these is actually really cool, where they don't just feed Claude information, they can actually open up live right inside of the chat.

12:25So I can just ask different things about my projects, and it could pull my actual Asana or ClickUp board into the conversation, and I can just check things off without ever having to leave Claude. Now if there's any applications that you use inside of your operation that are not currently available, you can use something like Zapier.

12:41So a Zapier connection is going to be able to reach thousands of more integrations on its own. You can just plug your own on top of that.

12:47So whatever you use day to day, CloudKit almost certainly gets to it. How you would actually do that is you go to add custom connector, and you're just going to be using MCP. So for example, if we are using Zapier, then we can just go to Zapier MCP, and we'll just have to connect a couple of different links.

13:05So I have a link down below in the description to Zapier, and just go to create a new MCP server, and you'll click on connect, and it gives you the instructions of how you can actually connect this back into Clot. Extremely simple. It's about two different steps.

13:15And when it comes to actually wiring any one of these up, let me show you. So for example, if we go to something like Canva, it's typically just using an OAuth where you can click on connect. It's then going to just prompt the window for you.

13:27You log in to your account, it's that simple. I click on allow, we now have our authorization as a success. We'll open Claude back up, close this out, and we should be good.

13:38Just like that, you can see we now have Canva as our connector, and we can determine what we want to allow it access to. So giving it access to write and delete tools, we can make sure it's always needing approval or always allow things or we can block certain things or even customize and give it a specific instruction for what we want it to give it access to.

13:57Now for you, some of the most important ones that you'll probably wanna be connecting is your email. So either Gmail or maybe OneDrive giving an access to Microsoft and that whole suite. But Gmail, Google Calendar, extremely simple to set up.

14:09You just log in to your account similarly to what I just showed you with Canva. So given we already have all of these connectors, let's put some of them into place. Let's go back into the chat section.

14:17We'll just start off new chat. I'm gonna ask look at my calendar for this week and tell me what I should prepare for and which days are going to be tight. Right off the bat, it'll pull up my calendar for the week.

14:25Let me load the tools first. So it's going to find the tools that it has available to it, and then it's going to pull this week's events, looking at this week starting today, and mapping out the schedule by day. So I'll note the recurring items, like the morning routine and Wednesday church service, but focus on the specific appointments, starting with today's office hours with Nick on Google Meet at eleven.

14:45So it's pulling every single individual item, addressing it one by one. And just like that, here's how your week shapes up. So we have today, we have Wednesday, looks like this is your tight day, and then we have Thursday, and Friday, and of course our weekend as well.

14:59Level four, and this is the one that made me fall for Claude in the first place, it doesn't just write text back to you. So it can actually build things, and they are called artifacts. So an artifact, this is more or less just a finished thing that is going to pop up in its own panel next to your chat.

15:15So as I showed you earlier, that vacuum document, it's now sitting in this artifact section. So it just show a visual representation of the document on the right hand side.

15:25So it can be a document, it can be a chart, it can be a little website, or a working interactive application, and you can just keep shaping it just by talking to it. So let me build one so you can see more specifically what I mean.

15:37So I'll go back into the new chat, and this time we're going to ask it to build me an interactive mortgage calculator. So I put in the loan amount, the interest rate, and the term, and it shows my monthly payment, plus a clean chart of how much interest I pay over the life of the loan.

15:50Keep the design simple and modern. Just like that, as it's finishing up, you can see it is prompting on the right hand side, an HTML page. So we have the mortgage rate here.

15:57We even have this interactive slider where we can play around with the loan amount, the interest rate, and the terms, so it can obviously look a lot different, and customize it as much as we need. Now anything beyond this, if we need to tweak just one small thing, we can easily just go and type in plain English what we want changed, we could take a screenshot of anything.

16:16But there is actual better ways to be, you know, annotating and customizing anything inside of your HTML. I'll show you that a little bit later.

16:24Again, every artifact that you create is then going to be living inside of this artifacts tab right here, so I have quite a long list of artifacts. Next up, we're gonna ask for something a little bit differently. I'm gonna ask to build me a quiz application.

16:35I tip in any topic, and it generates five questions one at a time, checks my answers as I go, and at the end, it gives me a score and tells me what to study, make it look clean and fun. So the reason I'm doing something else this time is because I wanna show you something specific to where you can actually deploy this live.

16:50So like that, we now have our quiz on the right hand side. We can ask it any questions. So I'll just stay I'll actually take one of these.

16:56So we'll say the Roman Empire difficulty easy, start quiz, and it should generate that and spin it up in just a second. Here we go.

17:02Question one, what was the capital city of the Roman Empire? And we have all these different options right here. Click on Rome, check answer, see the correct answer or if we guessed wrong, and then it'll just take us to question two and from there on.

17:14Now one of the best parts about this is if we go to this tab right up here at the top, we can click on publish artifact. So what we can actually do is now share this with anybody else. So if we copy this link, we can now go into the web, and let's pull this up so we can share it with anybody else who wants to take this quiz.

17:30So I can share it with friends or share anything that I create for my team, and they can now use it as well. So you just built and shipped a little AI application with just about one sentence.

17:39Level five, this is the part that beginners usually skip, and it's what took Claude from something that I used to something that kind of rents itself for me. And that is teaching it your own way of doing things. So it all is going to be living inside of here, inside of this customized section.

17:54So there's a few things in here, but the two that I really want you to see are these. It's going to be the skills section specifically, and then the plugins from there.

18:03So the simplest way to think about them is that a skill it's just going to be a recipe, one task that you have taught Claude to do in your way, so you are never having to re explain it. And a plug in, a plug in is really just going to be a playbook. So it's a whole bundle of all of those recipes with the applications and the tools that they already need wired in in one click.

18:25So recipes and playbooks, that is it.

18:28Now let me first just create a skill because it's actually way easier than it sounds, you don't need to write any code. So say there's something that I do over and over again, one specific process, and for me, maybe it's just turning messy notes into something useful. So what I can do is I can simply just ask Claude to help me create a skill where when I paste some messy notes, turn them into a clean summary, a list of action items, and any open questions, always using the same format.

18:53We'll round this off, and it should spin up the skill creation skill. So right here, it's going to be reading the skill creator skill to follow the right structure, and if we actually dive into that a little bit further.

19:03So if we go to the skills, we can find that there's a skill that is just by default implemented into Cloud, and as you would have guessed it, it's just for creating specific skills.

19:14So that is exactly what it is doing right now. We have other ones like an LLM console, where we ask one AI a question, you get one answer. And then we have other different skills like an invoice generator, where it's just able to provide invoices in my name, content strategy, CEO advisor, financial analyst, follow-up emails, content repurposing, a humanizer where it just makes it not sound like AI slap whenever we're writing emails or creating messages back to any people.

19:42And just like that, we can see we are now drafting the notes cleanup skill.md. It's gonna be refining everything and creating that markdown file for us because that is fundamentally what this is.

19:52What a skill is, it's just a markdown file in YAML format. And just like that, we now have our skill created. It's automatically going to pull up the artifact for us.

20:00And we can see right here. So we have, you know, the specific directions of this output format, and it's just gonna be adhering to the structure that, you know, the skill creator says it needs to be in specifically.

20:11So before we start testing it out, let's first just make sure that this is ready to go. So I'll say confirmed, now publish.

20:18And just like that, we now have everything, so we can click on save skill. It's now saved, and if we go into our customize section, we should see it inside of our skills section right here. Just like that, notes cleanup.

20:28And if we wanted to actually utilize it, we can just do slash, and we should have the notes cleanup just right there. So either we can call this directly or we can insinuate to use it without actually having to do the slash command, but either approach is going to work for you.

20:41Now let's discuss the plugins. So plug ins, it's just a bunch of those skills bundled up for a whole kind of work. So there is actually going to be a directory for them.

20:51If we go to browse plug ins, we can see all of them right here. So we have small business. So if we click on any one of these, it's just gonna be a bunch of different processes that small businesses are typically doing.

21:02So we have all the most common connectors. We have Stripe, Slack, DocuSign, Canva, HubSpot, PayPal, QuickBooks, and a few others. And it's not just gonna be limited to this.

21:11You can actually add your own connectors as well, and then all the different skills. So contract review, so being able to review all contracts. A Monday brief, where it's going to generate a one page Monday morning brief.

21:22We have a let's see another one. We have a closed month, so helping you close your books for the month, plan your payroll, onboarding any clients, helping with that entire process.

21:31Let's back up and go into some other ones. So let's go to let's see.

21:36Maybe just sales. So this one, as it says, it's prospecting, crafting outreach, and building deal strategy faster, so it can prep for calls, manage your pipeline, and write personalized messages that moves deals forward. So here's some of the skills that are embedded within this, and the coinciding connectors alongside it.

21:55Now skills, they're going to be free for everybody in building your own and using them. It all works on a free plan, and the only paid part here, it is going to be the plugins right here. So those bundled playbooks.

22:05So to install one of those, you'll just want to be on pro, but making your own skills like I just did, that's totally free. To add something like this yourself, you can just click on any one of them, and you click on install. So I already have this one installed here.

22:16If you go to Datadog, you just simply click install. It'll go into your section right here.

22:21You can see all the ones that I currently have access to. We can play around with any one of them. We can customize any of them as much as we would like.

22:27We just have to type into plain English and just be specific about what we want done. Okay. Now I do wanna cover these last two tabs, co work and Claude code, so you know what they are when you actually see them.

22:38So this one right here, we have Claude Cowork. And in Thropic, they basically just built this for people who are not coders. So any founders, business owners, and anyone who actually wanted to use Claude to actually do the work, but had no interest in touching a terminal.

22:51So in the chat, Claude is, of course, going to be talking to you, where in Cowork, it goes and does the work on its own, more in-depth, inside of a folder that you can be pointing it to right here, you can see all of my folders. This is everything on my local machine right here. So the idea is you hand it a task, like maybe clean up a messy folder, or pull a report together out of like the files on my computer, and it just goes and it does it.

23:16So it checks in with you as it works, and it always asks for changes or deletes anything. And beyond that, it can do a couple of things that chat just simply cannot. So you can set it to run on a schedule, so it does the same job for you every single morning or whenever you specify.

23:32It can build you live dashboards that just keep constantly updating themselves, and you can even kick a task off from your phone, and it'll just go work on it back at your computer. It's really the closest thing to Claude sitting down at your desk and just handling something entirely for you. And then lastly, there is going to be Claude code.

23:50So this is the third way to be using Claude, and the most powerful one. So this is Claude just built for actual, usually, shipping software where it can read your whole code base, edit code across a bunch of different files, and you just have to review the changes and approve them.

24:06So it's gonna be working just like a coding teammate that does the heavy lifting for you. Now if you're not a developer, I would still be recommending Claude Code over Claude Cowork.

24:15It is just going to be much better overall, and it has just some more features that Claude Coburg does not offer. Because again, like the level to build anything, like the floor has just dropped significantly, so you really don't need any technical expertise or prowess.

24:28Me personally, I have zero technical background. I didn't graduate with a CS degree or anything like that, and I used Claude Code for about 99% of my work. Now I do have full videos on Claude Cowork and Claude Code, so make sure to check those out on my YouTube channel.

24:41I also have some more in-depth ones that I don't upload on YouTube inside of my free community. So if you are interested in that, go check that out, link down below in the description. And one of the last things that I do want us to be covering is going to be the models.

24:52So if we just go back into our chat, we can really go into any of these tabs here, it doesn't really matter, you can see we have a list of different models to be choosing from. So at the top, it is going to be Opus 4.8.

25:03This is their flagship. It's brand new. It really just came out, and it's the most capable model Anthropic has ever made, or at least released to the public, because if you are familiar with Claude Mythos, that is apparently like the best of the best.

25:16But anyways, this is going to be the best at, you know, the hard thinking, the writing, and the building. Now one thing to know is that Opus, it is part of the paid plan. So if you're on pro or above, this is your default, and you can leave it right here for almost everything.

25:30Under that is going to be Sonnet. Now this is the quicker one. It's still very capable.

25:35It's just faster and lighter on your usage, and this is what you're mostly on with the free plan. So if that's you, you're in good shape. But even when you're paying, dropping to Sonnet for lighter, just back and forth, is a nice way to stretch your limits, so you get more usage and you're not just using all of your tokens using Opus 4.8.

25:52Because you don't need to, you know, have the best reasoning and use all these different tokens for very simple tasks as you would imagine. Below that is going to be Haiku. So this is, you know, like the little model, and it's built purely for speed.

26:05So it's the fastest, it's the cheapest, it's made for very quick simple tasks where you just want the answer pretty much instantly. So you won't really be reaching for it often unless really depending on your work, of course. But when you want speed over depth, this is going to be the best model for you.

26:19Now next to all of these models, you will be able to see a little setting that we have here, and this is just going to be a little slider where this is just the effort. So it's just how hard Claude thinks before it answers. So by default, it is going to be sitting on high.

26:34So right now I actually have it as max effort, and there's ultra code, which is, as you can imagine, it is going to be very strong. But we have high all the way downs to low. So if you're handing Claude and assigning it something pretty gnarly and you want it to go deeper, you bump it up to one of these top ones like Max or UltraCode.

26:51But the usage, it is going to be consuming much higher rates than anything lower than that as you would expect. So that's the full picture, start to finish.

26:59And if you guys do take just one thing from this entire video, take the projects. So maybe spend about ten minutes tonight, set one up, and you can just drop in a quick page about yourself, and let Claude get to know you. Everything else that we just went through, it works better once you've actually done that, and you you have the fundamentals of understanding Claude.

27:16But that's where I would start if you're a complete beginner, and that's gonna do it for this one. So if you guys made it all the way to the end, thank you guys genuinely. Make sure to check out our free Skull community for all the guides, all the resources.

27:27Link will be down below in the description. And if you guys wanna take it a step further, make sure to check out all our other videos on quad code, quad co work. That's going to be the next step.

27:36But in any case, thank you guys for watching. I'll see you in the next video.

The Hook

The bait, then the rug-pull.

The opening line does the work most tutorials bury in a minute of preamble: Claude is not a search box. Everything that follows — five levels, sixteen chapters, twenty-seven minutes — is the proof.

Frameworks