Stop Watching Tutorials - Build These 4 Claude Projects to 10x Output
By Austin Marchese
Summary
Topics Covered
- Clone Your Personal Board of Advisers
- Build for Yourself, Skip Audience Pressure
- AI Is Now Your First Online Impression
- Your AI System Should Improve Itself
Full Transcript
Project number one is your own board of adviserss. This project will let you
adviserss. This project will let you build your own board of adviserss by cloning professional experts so you can make decisions with confidence. So step
one here is let Claude interview you.
Run this prompt so Claude can get to know your goals and where you want to go with your career. The key phrase here is the interview me part which tells Claude to pull context from you instead of guessing or making assumptions. And the
pro tip here when you're answering these questions is to use a voicetoext tool like call it whisperflow or hex which is the free open- source alternative which will allow you to talk instead of typing which is like 10 times faster. Step two
is identify your board members. Once
Claude knows what you want to accomplish in the next 1 5 10 years, ask it to suggest who should be on your board. For
this video, we're going to start with two people, but you can add as many as you want. Generally, I like to bias
you want. Generally, I like to bias towards people who have publicly available information. for example,
available information. for example, YouTube creators because their full body of work is accessible and that serves as training data as you recreate these people on your board. And for any prompt that I go through, just take a
screenshot of it and then send it into Claude Code. It's essentially the same
Claude Code. It's essentially the same as typing it. So, after you know who's on your board, step three is ingest their content to create their clone. So,
go to YouTube or wherever that person has training data available and pull all the links to the content that you want to ingest. Then paste that right into
to ingest. Then paste that right into Claude and ask Claude to ingest that data into your project. Here you can see a prompt I use which helps me ingest specifically YouTube data. Step four is create your board skill. Once steps one
and three are done, create a custom skill that lets you ask your board questions in one go. We'll go deeper on the concept of custom skills in Project 4, but the quick version is that a skill lets you run a multi-step process
automatically by just calling that specific skill. If I had to really
specific skill. If I had to really simplify it, just think of it as a reusable prompt that you can easily update. In this case, we're going to
update. In this case, we're going to call the skill ask the board. So Claude
knows to loop through every single board member instead of just one and then gets you a response based on all of their analysis. Step five, now that you've
analysis. Step five, now that you've created the skill, you can actually go ahead and ask your question. Just type / ask the board followed by whatever is on your mind. So for example, it could be
your mind. So for example, it could be based on everything you know about me, what should I do to capitalize on AI?
When I was a COO of my last tech startup that raised over $18 million, I use this exact process to help me make any big decision that I was doing. I can't
stress how valuable this one project is, but that's just the first of four Claude Code projects you need to do, whether you're technical or not. And I promise if you do all four of these, you will walk away feeling 10 times more
confident about building with AI.
Project number two is create your niched command center. The goal of this project
command center. The goal of this project is to get your hands dirty building a tool that helps you today with something you're already working on. For me, I spent a lot of time on YouTube, so I built a YouTube planning command center.
And before we get into how to build your niche command setter, there are four reasons why building this is so valuable. The first is that you'll
valuable. The first is that you'll actually use it. Most tools you'll create just sit unused because they solve hypothetical problems. This is designed to solve problems you have today. Two is you skip the hardest part.
today. Two is you skip the hardest part.
Coming up with the right problem is what causes analysis paralysis and by building it for yourself, this skips all of it. Three is you get sharper at the
of it. Three is you get sharper at the thing you actually care about. Building
this project forces you to map your workflow and determine how you actually do things. and the tool is nice that
do things. and the tool is nice that you'll be creating, but flexing your critical thinking muscles is the massive unlock. And the fourth and probably the
unlock. And the fourth and probably the most important is that there is zero audience pressure. You don't have to
audience pressure. You don't have to ask, "Does this look good?" You don't have to make it pretty. You don't have to think about marketing. It's literally
just for you. So, you could build quickly. So, step one of this whole
quickly. So, step one of this whole process is choose what to build. If you
don't know what to build for your niched command center, this prompt helps you create a list of options based on your previous chats with Claude. pick one of those concepts. But for this video, I'm
those concepts. But for this video, I'm building a personal finance tracker that charts all of my expenses and income.
And before we get to step two of this project, the question that's probably coming up in your head is, how can I make this so anybody can use it if I want? We'll show that in project three.
want? We'll show that in project three.
And what makes it 10 times easier is today's sponsor. So, shout out to the
today's sponsor. So, shout out to the goats at Hostinger. Hostinger just
launched a hosting plan built specifically for Claude Code users. And
amongst the premium services out there, it is by far the most affordable starting at as little as $3.99 a month, which is less than a cup of coffee. But
the part that I'm most hyped about is the Hostinger connector. It's an MCP server for Claude Code. To deploy an app that you're working on, you just need their API key and then you bring that into Claude Code and it can deploy the app right from your terminal. You build
in Claude Code, you deploy in Claude Code. One session, you essentially don't
Code. One session, you essentially don't ever have to think about it ever again.
and let's say something breaks, their AI troubleshooter can scan the logs and tell you exactly how to fix it. In
Project 3, I'll be using this exact feature to deploy an app that we build.
And for that, I won't have to buy a domain because Hostinger gives everyone access to select domains for free for the first year. So, you'll see me using this tool in a bit, and I've used this for years. So, to get access, click the
for years. So, to get access, click the first link in the description, then use code Austin M to get 10% off. Now, let's
go back to your niche command center.
Step two of this is you want to plan and build out the MVP. There is a universally agreed upon best practice when using tools like Claude Code. You
want to plan before you build anything complex. So step back, breathe, and
complex. So step back, breathe, and think about what you need Claude to build before it actually builds it. The
key here is a planning prompt. This is
similar to project one where you let Claude interview you, but there's technical guardrails baked in. This
interview will help you build a plan for the features you actually need, not what some template online says about a finance dashboard or something. Then
after going back and forth and landing on a plan that you like, you can improve it and let Claude rip. And if you hit any issues, just copy and paste the error that you see into Claude or screenshot the error and just fire it
in. Once Claude creates this minimum
in. Once Claude creates this minimum viable product that you can start using, just start using it. The point isn't to launch something perfect. It's to create something that's valuable today and add features as you need them. For the
personal finance example, let's say it doesn't look at your credit card transactions. You could just say, "Hey,
transactions. You could just say, "Hey, I want to add credit card transactions to it." and just continue this iteration
to it." and just continue this iteration cycle as quickly as possible and your niche command center is going to start becoming extremely valuable. I built the finance tracker as an example I knew would apply to most of you, but the one that I actually use is this YouTube
tracking dashboard on screen. At this
point, it's pretty robust and it has all the features I need to help plan my YouTube videos. And now I don't have to
YouTube videos. And now I don't have to pay for any external subscriptions.
That's a added bonus. And every single feature is specific to exactly what I need. If somebody else tried using this,
need. If somebody else tried using this, it likely wouldn't work that well. So,
this niched command center is built just for you. But what if you want other
for you. But what if you want other people to start using it? That goes to project number three, which is building your AI optimized public profile. For
this project, we're going to create a personal website that is optimized for AI SEO. There are two goals here. First,
AI SEO. There are two goals here. First,
humans land on it and they think, "Wow, this is awesome." And they get a great first impression of you. Personally,
I've always loved building websites. I
built awesomercasy.co years ago. It's the first engineering
years ago. It's the first engineering project I ever built. Just because I believe so much that you should just have your personal website. It's fun.
It's creative. if it's awesome. Second
is that people will start going to AI first to learn about a person. When
someone types in who is awesome Arcesy, I want my personal website to help tell the story. So to do this, step one is
the story. So to do this, step one is set the goal of the page before you build anything. The question you want to
build anything. The question you want to ask yourself is when someone lands on your page, what do you want them to think or what do you want them to walk away with? For me, this is somebody
away with? For me, this is somebody who's smart. I want to work with them.
who's smart. I want to work with them.
Pick what you want and then reverse engineer the styling and the structure based on that feeling. Step two is pull design inspiration. Find sites that you
design inspiration. Find sites that you like, a creator you admire, a brand that you think is dope. Just find things that you love and that can be a source of inspiration. Or you could go to any of
inspiration. Or you could go to any of these websites and they have a bunch of inspirations already pre-curated so that you could just find something that you love. Step three is create a NodeJS app
love. Step three is create a NodeJS app with Claude code. So what you want to do is you take the goal from step one, the inspiration from step two, and then have Claude scaffold an initial proof of concept. This prompt will help you
concept. This prompt will help you create it and interview you to fill in the copy. And in the prompt itself, it
the copy. And in the prompt itself, it says, "Build me a personal website in Node.js. JS that'll help us deploy it to
Node.js. JS that'll help us deploy it to Hostinger in a little bit. Step four is optimize for AI. We're starting to see the trend where people aren't going to Google to find answers, they're going to AI. And so that means AI is likely going
AI. And so that means AI is likely going to be the first that looks you up. And
whether that's a recruiter that AI scans about you or a buyer looking to purchase from you or teammates looking to see who the hell is this guy? Basically, anyone
trying to figure out who you are. So to
optimize this, we're going to have Claude generate these following things.
And then at the bottom of each page, we're going to add an ask AI about me block, which will have links to each of the AI providers. Then when someone clicks that button, it'll go to the AI chat with a preloaded prompt, and when
they hit enter, it'll load information about you. I love this little feature,
about you. I love this little feature, and I think it's going to be part of most modern websites at the footer, the same way social links are there, but it essentially sets you up for success for a person to go into AI and ask it about
you. Send this prompt into Claude to
you. Send this prompt into Claude to help set it up. At this point, you have your AI optimized website. And now step five is deploy the app. There are a number of ways to do this, but today we're going to use the Hostinger connector. To do this, go to their
connector. To do this, go to their website and then pull the API key from there. And then we're going to use that
there. And then we're going to use that to configure the connector. You can have Claude do all of this for you once you actually have the API key and the connection string or you can edit the MCP file directly. After that, I select the free domain that I want to use for
this site. For me, I'm going to deploy
this site. For me, I'm going to deploy to awesome case.te. Then in your Claude code session, you can send this prompt and that's that. Claude handles the deploy and it's really that simple. And
if anything fails, Hostinger's AI troubleshooter will scan the logs and tell Claude what's wrong. This makes it so Claude can fix it in seconds. And
there are other ways to do this, like using GitHub to autodeploy through Hostinger, which is likely a better option for a fully scaled out product, but for the sake of this project, that's not needed. Now, at this point of the
not needed. Now, at this point of the video, you've created your own board of adviserss to help you get highquality answers. You've created your niched
answers. You've created your niched dashboard, and then you've created your own personal AI optimized website. Now,
project 4 is the one project that made me go from AI is awesome to AI will change the world forever. Before we get into that, if this is your first video, welcome the channel. But if it's your
second or more, here is our anti-slop agreement. The visuals, the testing, the
agreement. The visuals, the testing, the time I put into this, this is for humans, not for AI clanker robots. So,
all I ask is that you subscribe as part of this agreement to help this content reach more people. Also, I do want to congratulate Ali Nazavar 6127 for winning our Claude Max giveaway. Comment
below with what you're building to enter the next giveaway. Now, project 4 is about creating your own internal operating system. This is a project that
operating system. This is a project that personally made me realize or confirm that AI was going to change the world.
So, your advisor board, your command center, your public profile, these are all tools you built. The internal
operating system is the system that can hold all of it together. Simply put,
it's a bunch of files that help AI go from a good output to a hyperspecific great output. And there are three steps
great output. And there are three steps to get this set up. The structure, the improvement loop, and the skills that maintain it. And I use this for my own
maintain it. And I use this for my own team and I've set this up for other seven figureure businesses and every time I do it, they're just like, "Yo, this is crazy." And I'm just like, "I know." Step one is set up the structure.
know." Step one is set up the structure.
There are three folders in this system.
Knowledge is everything you want Claude to know. This is meeting notes, voice
to know. This is meeting notes, voice samples, frameworks you trust, articles you've saved. This is where your board
you've saved. This is where your board of advisers content would live. There's
skills. This is your repeatable processes. The ask the board skill we
processes. The ask the board skill we created in project one would live here.
And then there's project. This is what you're actively working on. So, for
example, your command center and your public profile could live here once you start building them out. Here's a prompt you can run to set up this structure.
There's a couple of keys here, but one of them is that claude.md is the file at the root that is the brain of the whole thing. It essentially tells Claude how
thing. It essentially tells Claude how to use the folders, so you don't have to reexplain every time you open a new session. It's like teaching Claude how
session. It's like teaching Claude how to use the system. Step two is you want to build a slashimprove system skill.
This is the most important part of the whole process. Most people use AI, they
whole process. Most people use AI, they get an output that's okay and then they just accept it. And I frankly don't blame them because most of us have never had to manage or train someone to upskill them. So it just doesn't really
upskill them. So it just doesn't really come naturally. So the key with AI is
come naturally. So the key with AI is creating a system that logs information, which we did in step one, and then a system that captures feedback so your system improves over time. That's what
the slashimprove system skill does.
Here's how it could work. Let's say you write an email and the first draft is too verbose. You could say rewrite
too verbose. You could say rewrite shorter and eventually you land on a good draft. Then after landing on that
good draft. Then after landing on that final draft, you would run /improve system. And then the next time it writes
system. And then the next time it writes an email, it'll automatically be more concise. Every iteration, it teaches the
concise. Every iteration, it teaches the system and it improves. On screen, you can see a simple version of this improve system skill that you can immediately set up. If you want a more productized
set up. If you want a more productized version of this that I use for every client I work with, if you go to buildpartner.ai, it has a Claude Code plugin that does exactly this, and it's free to try out. That version allows you to run it once, and it'll audit your
whole setup, suggest improvements, and it'll update files for you. Step three
is build an ingest resource skill.
Improve system is a teacher that helps your internal OS learn. The ingest
resource skill lets you systematically bring in new information, call it articles, transcripts, YouTube videos, and ingest it into your system to make sure it's set up properly. Think of this like a librarian that helps you get information and then put it in the right
place so that the AI brain knows where to look. Step four is you put it all on
to look. Step four is you put it all on GitHub. And this is technically
GitHub. And this is technically optional, especially if you're just using this yourself, but I do think it matters. I think of my internal
matters. I think of my internal operating system as the intellectual property of my life. And it started with just me using it, but now my whole team is plugged into it. And every time I set up something similar with my clients, they're just blown away. For this
reason, I like to treat it as software.
And generally speaking, software lives on GitHub. And by adding it, it's easy
on GitHub. And by adding it, it's easy to undo any changes that you make and you can see your whole progress over time. Then eventually, it sets it up so
time. Then eventually, it sets it up so anybody can start using this system.
Here's a prompt you can use to help you get started with this. The key with this internal operating system is that you create a foundation that can build up over time. Data ingestion that allows
over time. Data ingestion that allows you to build up a valuable data set and a simple way to improve the system. So
once you create your board of adviserss, you set up your niche command center, you build your AI optimized public profile, and then you set up your internal operating system, you will have an entirely different swagger about building with AI. My advice is just
start doing things. Don't overthink it.
Now, if you want to go deeper on the internal operating system, I have this video where I break down how Andrea Carpathy, the former head of AI at Tesla, set up his own version of this.
It builds on everything that we covered in project 4 and gives an in-depth tutorial on how to do this. I'll see you over there. Peace.
over there. Peace.
Loading video analysis...