You've been using AI Wrong

Summary

The video introduces Fabric, an open-source AI tool designed to augment human capabilities by simplifying AI interaction. Creator Daniel Meer discusses reducing friction for users to leverage AI for problem-solving. Viewers learn to set up Fabric, use built-in 'patterns' for tasks like extracting insights from videos, and even create custom patterns. The tool's potential for personal and professional augmentation is highlighted, emphasizing its role in helping humans flourish by managing information overload and enhancing productivity.

Takeaways

• 🤖 Fabric is an open-source AI tool designed to augment human capabilities by reducing the friction to use AI for problem-solving.
• 🛠️ The primary goal of Fabric is to make AI more accessible and easier to use, with a focus on solving specific problems through 'patterns', which are curated sets of instructions for AI.
• 🔍 Fabric allows users to extract insights and summaries from various forms of content, such as YouTube video transcripts, by using built-in tools and patterns like 'Extract Wisdom'.
• 🌐 The tool is CLI (Command Line Interface) native, enabling users to interact with AI directly from the command line, which is beneficial for those proficient with terminal usage.
• 💡 Fabric's 'patterns' are open-source and crowdsourced, meaning they are continuously improved by the community, and users can also create their own custom patterns.
• 🔗 The tool can be integrated with local AI models or cloud-based APIs from providers like OpenAI, Anthropic, or others, requiring API keys for access.
• 🌟 A unique feature of Fabric is the ability to 'stitch' patterns together, allowing for a sequence of AI processing steps to be executed in a pipeline.
• 🔑 Users need to set up their API keys for various services within Fabric to utilize its full capabilities, including accessing local and remote AI models.
• 📝 Fabric supports saving outputs directly to note-taking applications like Obsidian, enhancing the workflow for users who rely on a 'world of text' for information management.
• 🌱 The philosophy behind Fabric is to aid in human flourishing by helping users identify, articulate, and pursue their purpose in life through AI augmentation.
• 🚀 The tool is in active development with new features being added, such as the ability to save outputs to Obsidian and improvements in pattern creation and management.

Q & A

What is Fabric and what is its primary goal?

Fabric is an open-source AI tool created by Daniel Meer, designed to augment humans with AI capabilities. Its primary goal is to reduce friction in using AI to solve problems, making AI more accessible and easier to integrate into daily tasks.

How does Fabric help in reducing friction for using AI?

Fabric reduces friction by providing a command-line interface (CLI) that allows users to interact with AI through pre-defined 'patterns' or prompts. These patterns are designed to solve specific problems and can be easily integrated into existing workflows.

What is a 'pattern' in the context of Fabric?

In Fabric, a 'pattern' is a set of instructions or a way to get the AI to perform a specific task. It's essentially a prompt that has been carefully curated and crowdsourced to elicit the best response from an AI for a particular problem.

How does Fabric handle the use of AI models from different platforms?

Fabric is compatible with various AI models from platforms like Open AI, Anthropic, and even local models with Alamaze. It sends text to the user's preferred AI model and processes the response according to the selected pattern.

What is the 'Extract Wisdom' pattern in Fabric and how is it used?

The 'Extract Wisdom' pattern is a feature within Fabric that processes text, such as a YouTube transcript, to extract key insights, quotes, and ideas. It's designed to quickly provide users with the essential takeaways from lengthy content.

How does Fabric allow users to create their own patterns?

Fabric enables users to create custom patterns by modifying existing ones or starting from scratch. Users can save their custom patterns in a local directory and integrate them into Fabric for use in their workflows.

What is the significance of the 'World of Text' concept mentioned in the script?

The 'World of Text' concept refers to the practice of capturing all information in text format, making it easily manipulable and accessible by AI tools. This approach allows for better organization and retrieval of information, as well as enhanced AI integration.

How does Fabric facilitate the integration of AI into programming and data handling?

Fabric allows users to incorporate AI capabilities directly into their scripts or programs. For example, it can process JSON data from an API or interact with local AI models to perform tasks, reducing the need for complex API interactions.

What is Twin Gate and how does it relate to Fabric?

Twin Gate is a remote access solution that allows users to connect to their home or office servers from anywhere. It's mentioned in the context of accessing a local AI server for Fabric when the user is away from their main setup.

How can Fabric be used to enhance personal productivity and knowledge acquisition?

Fabric can be used to summarize and analyze content, filter out what is worth in-depth attention, and even record and transcribe conversations for later review. This helps in efficiently managing the consumption of information and enhancing personal productivity.

What is the potential impact of Fabric on the way humans interact with and consume content?

Fabric has the potential to significantly change content consumption by enabling users to quickly extract valuable insights from vast amounts of information. It promotes a more efficient and focused approach to learning and staying informed.

Outlines

00:00

🤖 Discovering Fabric: The AI Tool for Human Augmentation

The speaker introduces Fabric, an open-source AI tool created by Daniel Meer, designed to augment humans with AI capabilities. The focus is on reducing friction for utilizing AI to solve problems. The speaker shares their daily use of Fabric and previews a demonstration of setting it up and using it to access a local AI server. The tool's ability to extract wisdom from a lengthy YouTube video is highlighted, showcasing its efficiency. Fabric operates by sending text to AI models and uses 'Extract Wisdom', a pattern or set of prompts, to guide AI responses. The concept of patterns is discussed as an open-source, crowdsourced method of refining AI prompts to solve specific problems effectively.

05:01

🔧 Setting Up Fabric and Embracing the World of Text

This section provides a step-by-step guide on setting up Fabric on various operating systems, including Linux, Mac, and Windows via WSL. The process involves cloning the Fabric project from GitHub, installing PIP x, and using it to install Fabric. The setup includes updating system repositories and ensuring path variables are correctly set. The speaker also introduces the concept of a 'World of Text', emphasizing the importance of capturing and manipulating text for AI processing. The video also mentions the need for API keys from Open AI, Anthropic, and YouTube to utilize Fabric's full capabilities, acknowledging that while Fabric is a framework, it relies on external AI services which may incur costs.

10:04

🌐 Accessing Local AI Models and Remote Servers with Fabric

The speaker discusses the capability of Fabric to work with local large language models (LLMs) and how it can be used to summarize content from sources like YouTube videos. They demonstrate using Fabric with a local model called LAMA and show how to switch between different models. The integration with Twin Gate, a remote access solution, is highlighted, allowing the user to connect to a remote AI server from anywhere. The setup for Twin Gate is briefly explained, showcasing its ease of use and the ability to control access to specific resources.

15:04

📚 Advanced Fabric Features: Stitching, Context, and Custom Patterns

Advanced features of Fabric are explored, such as stitching, which allows combining multiple patterns to perform complex tasks. The speaker demonstrates summarizing an article and then using the summary to generate an essay. They also discuss the ability to analyze claims and rate the quality of content. The concept of context in Fabric is introduced, which helps define the user's goals and intentions for using AI. Additionally, the process of creating custom patterns in Fabric is outlined, emphasizing the user's ability to solve specific problems by crafting and iterating on their own patterns.

20:05

💡 The Philosophy Behind Fabric: Augmenting Human Intelligence

The underlying philosophy of Fabric is discussed, focusing on its aim to augment human intelligence rather than replace it. Daniel Mesler's background as a hacker and his decision to go independent upon the release of GPT-4 are shared. The speaker talks about the importance of processing and thinking in personal growth and how Fabric can be used to filter and prioritize content consumption. The tool is positioned as a means to help users determine which content is worth their time and attention, emulating the process of manual note-taking to enhance understanding and retention.

25:06

🔗 Integrating Fabric with Obsidian: Saving AI Insights

The final paragraph covers the integration of Fabric with Obsidian, a note-taking application. The speaker demonstrates how to save the output from Fabric directly into Obsidian, creating a new note. This feature is particularly appealing for users who maintain a 'world of text' and wish to incorporate AI-generated insights into their笔记 system seamlessly. The process involves setting the correct environment variables to ensure Fabric knows where to save the new markdown files in the Obsidian directory structure.

Keywords

💡AI

AI, or Artificial Intelligence, refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. In the context of the video, AI is central to the theme as it discusses the use of AI to augment human capabilities through the Fabric tool. The script mentions various AI models from companies like Open AI and Anthropic, illustrating the integration of AI into daily tasks to enhance productivity and knowledge extraction.

💡Open Source

Open source denotes a philosophy of software development where the source code is made available to the public, allowing anyone to view, use, modify, and distribute the software. In the video, the Fabric tool is described as open source, emphasizing its collaborative nature where the community can contribute to its development and improvement, as well as use it without restriction.

💡Fabric

Fabric is an open-source AI tool introduced in the video with the goal of augmenting humans with AI capabilities. It is designed to reduce friction in using AI to solve problems, making it more accessible and integrated into daily workflows. The script discusses how Fabric can be used to extract insights from videos, manage APIs, and even create custom patterns to solve specific problems.

💡Patterns

In the context of the video, patterns refer to a set of instructions or prompts engineered to guide AI in performing specific tasks. They are the 'secret sauce' behind Fabric, curated to solve particular problems effectively. The script highlights that these patterns are open source and crowdsourced, allowing for continuous improvement by the community.

💡Command Line Interface (CLI)

CLI is a user interface for interacting with computers where users input commands in the form of lines of text. The video emphasizes the CLI-native aspect of Fabric, which allows users to perform AI-related tasks directly from the command line, streamlining the process and making it more efficient. This is particularly appealing to users who prefer a direct and scriptable approach to problem-solving.

💡API Key

An API key is a unique code provided to users when they register for a service that requires authentication to access an API (Application Programming Interface). In the script, it is mentioned that Fabric requires API keys for services like Open AI, Anthropic, and YouTube to function, which connects the user's Fabric setup with these external services.

💡Transcription

Transcription is the process of converting spoken language into written form. The video script provides an example of using Fabric to transcribe a YouTube video and then extract valuable insights from it. This demonstrates the practical application of Fabric in processing and analyzing multimedia content.

💡Crowdsourcing

Crowdsourcing is a process that involves outsourcing tasks to a large, undefined group of people, typically via the internet. In the video, the development of Fabric's patterns is described as crowdsourced, meaning that the community contributes to their creation and refinement, leading to a more robust and diverse set of solutions.

💡Local AI Models

Local AI models refer to AI systems that run on the user's own hardware, as opposed to cloud-based services. The script mentions the option to use local models with Fabric, such as LAMA, which allows for processing data on a personal server, providing more control and potentially enhancing privacy.

💡Remote Access

Remote access allows users to connect to and control systems from different locations. The video discusses using Twin Gate for remote access to a local AI server, enabling the user to interact with their AI models even when they are not physically present near the server. This is particularly useful for maintaining productivity while working from various locations.

💡Human Augmentation

Human augmentation is the concept of enhancing human capabilities through technology. The video's main theme revolves around using Fabric to augment human intelligence and productivity. The tool is presented as a means to help users process information more effectively, make better decisions, and improve their overall ability to manage and utilize AI technologies.

Highlights

Introduction of Fabric, an open-source AI tool designed to augment humans with AI capabilities.

Fabric's purpose is to reduce friction for using AI to solve problems, as explained by its creator, Daniel Meer.

Demonstration of using Fabric to extract insights from a two-hour YouTube video transcript.

Explanation of 'Extract Wisdom', a pattern within Fabric that prompts AI to think deeply and extract key information.

Fabric's use of open-source and crowdsourced patterns to improve AI interaction.

The concept of patterns in Fabric as a set of instructions to guide AI in performing specific tasks.

How Fabric allows users to create and customize their own AI interaction patterns.

Fabric's Command Line Interface (CLI) and its benefits for reducing friction in AI problem-solving.

Integration of Fabric with local AI models and servers for on-the-go AI assistance.

Sponsorship mention of Twin Gate, a remote access solution for Fabric users.

The 'World of Text' concept by Daniel Mesler, emphasizing the importance of text for AI manipulation and human interaction.

Fabric's setup process for various operating systems including Linux, Mac, and Windows.

Instructions for installing Fabric using PIP x and setting up the environment for first-time use.

The necessity of API keys for using cloud-based AI models within Fabric.

Introduction of local LLMs in Fabric for AI usage without external data sharing.

How to use Twin Gate for remote access to local AI servers from anywhere.

Advanced Fabric features like stitching patterns together for complex AI tasks.

The ability to create custom patterns in Fabric for personalized AI assistance.

Philosophy behind Fabric focusing on human flourishing and augmentation rather than replacement.

Discussion on the importance of using AI to filter and prioritize content consumption for efficiency.

How Fabric can be used to analyze and store personal conversations for later reflection.

Integration of Fabric with Obsidian, a note-taking application, for seamless AI-generated note creation.

Transcripts

I found a new AI tool and I am obsessed.

It's completely open source and I use it every day. It's called Fabric. Daniel,

tell me what fabric is.

So basically the goal is to augment humans with ai,

so it's all about reducing that friction to be able to use AI for your.

Problems. That's Daniel Meer,

the creator of Fabric Reducing Friction so you can use AI to solve your

problems. That's the real purpose of technology and ai,

and that's what this Project Fabric is helping us do.

I legit use this every day and I think you might too.

So in this video we're going to break down what Fabric is and I'll show you how

to set this up so you can start using it right now and later in this video I'll

show you how I use Fabric to access my local AI server. Terry,

wherever I go with a sponsor of this video, twin Gate,

we'll talk more about them later.

Okay? You want to see something absolutely crazy?

Yes, Daniel, I do get you coffee ready? Let's do this.

Now before we dive too deep,

I want to give you a feel for what Using Fabric is kind of like,

what's the use case? What would I use this for?

This you're going to love watch this. Let's say I've got a YouTube video,

this two hour interview that maybe I don't have time to watch,

so I'll grab the link and on my command line I'll use the tool YT dash

transcript and paste the link,

and this is going to grab the transcript of that YouTube video. By the way,

this is a tool that's built into fabric. That alone makes this thing amazing.

Then I'll pipe that over into Fabric two hour YouTube video. Here we go.

Within moments I'm told about David Bumble and everything we discuss in this

video, the ideas, insights, quotes,

like this one man who said that smart guy.

So within a few moments I took a two hour YouTube interview and extracted all

the wisdom and insights I need to know. That's crazy, right?

So what's happening here? What is Fabric doing? Let's break that down.

First we start off with some kind of text. In this case, a YouTube transcript.

It really could be anything. And then Fabric will send this text, in this case,

our YouTube transcripts off to your favorite ai.

It could be models from open ai, anthropic or even local models with Alama,

which again,

I will show you how to access a local server from anywhere using to one gate

here in a bit. Now looking at the command,

it's not just sending the transcript by itself. Here you go, ai.

It's using this thing called Extract Wisdom. What is that?

This is the secret sauce behind fabric and something you should get pretty

excited about. Daniel, tell us more about it.

What I've done is take any piece of AI from any platform that is interesting and

usually the language that I care about is actually prompts.

So what we started doing is collecting all these prompts into this concept

called patterns.

A set of instructions or a way to get the AI to do what we want it to do. Now,

this is not a new concept. Prompt engineering is a thing, but this,

it's a bit different, different in two ways that I am a big fan of.

They're open source and they're crowdsourced. I'll be the first to admit.

That's kind of a weird concept for a prompt, but here's why it's cool. First,

these prompts or patterns, we'll just call 'em patterns from now on,

have been carefully curated, created, manipulated,

added to do exactly what it's designed to do to solve a problem,

a very specific problem,

and it's everyone including maybe you that helps create these prompts to make

them better. Now, let's take a look at this one in particular. I love this one,

extract wisdom. You'll really get a feel for what I'm talking about here.

And this is the other cool part about this.

It's open source so you can actually see this system prompt.

Normally when you're interacting with GPTs,

you can't see the actual prompt being sent to the AI here,

we're controlling that. We're part of that and I want to show you this one part.

I think this will illustrate what I'm talking about.

Look at how this prompt talks to the ai. Take a step back, think step by step,

think deeply. It kind of sounds like he's talking to this AI like a human,

and that's exactly the case. Even Daniel said.

You're basically telling it to act like a human.

We don't know why it works,

but just talking to these AI like they're humans elicits a better response,

better results. Kind of scary Do with that what you will,

but I point that out to say that these prompts have been tested time and time

again added to similar to what you would see with open source code,

and this is just one pattern. Look over here on the side,

look at all of these that have been created and you don't have to stop there.

You can create your own, which I've done that and oh my gosh, that's the secret.

It's so amazing. But hold on,

you might be like me and think this kind of seems just like fancy prompts.

What is this thing actually doing that I can't just do with chat?

GBTI want to revisit this idea Fabric is all about reducing friction to have

AI help you solve problems.

And one of the areas of friction I didn't even realize I had was the fact that I

had to keep going out to chat GBT, open up a web interface,

load up maybe a custom GBT or start having a conversation and it didn't feel

like a lot of time, but that is time it gets in the way fabric,

and this is one of the reasons I fell in love with it is CLI native.

You do everything here in the CLI, which I get may or may not excite you,

but you're not limited to that. Daniel Meer touches more on that.

What I'm trying to do is make the on-ramp to using these things as easy as

possible. So I want to be able to use them via voice.

I want to be able to use them via command line via a gooey app.

I want to be able to just access them as quickly as possible.

So that's the main thrust of this project is to collect

problems, collect the solutions in the form of these patterns,

and then to have as many on-ramps onto them as well.

Now I'll talk more about why I love that it's here in the command line,

but it's more than just how you interact with it.

Think about how you might use this to build programs instead of going through

the pain of interfacing with AI APIs.

So that's kind of hard to say actually it wasn't. Anyways,

you can just use fabric. Now check this out.

I want to show you something I did actually yesterday.

I'm trying to build my cardiovascular health.

So I started running and rowing and I tracked that with the app called Strava,

which gets all sorts of amazing data.

So I wrote a Python script to interact with their API to pull down all my data

and it looks like this, a bunch of messy JSON,

but I created a pattern called Workout summary and it takes care of the JSON for

me or I can just bake fabric right inside my Python script.

And that's just a simple example. This thing is crazy.

Now we're about to set this up, but I want to talk about one more thing.

It's called a World of Text and it's a concept that I'm

really adopting now, thank you to Daniel Mesler. Go ahead and tell him Daniel,

about this world of text.

20 Years ago I got into this guy named David Allen who basically said,

never ever store anything in your brain. Immediately capture.

What I do now is I capture a concept or a structure for an essay or something.

I capture it immediately in a note and now that it's text and because I'm fairly

proficient with Vim and the Terminal,

my whole world is text and the ability to manipulate text and I have all my

notes in text and when I record something that's actually sound,

I immediately transcribe it, send it to Notion, and so it's also in text.

So now I have this world of text that I could use and now I have this AI

infrastructure that manipulates text using AI to get results that help us as

humans.

So it's about getting everything into a text format so it can be used anywhere

by anything, especially ai.

And notice when I run these commands like getting this YouTube transcript,

it's outputting this in marked down format so it can play nice pretty much

wherever it goes, especially my notes application obsidian.

Now we'll touch more on the philosophy of why I think this is amazing and I'll

show you a few more patterns I've been working with.

But now let's get you set up. Let's get fabric on your computer right now.

And by the way,

if you want to see the full Daniel Meisler interview where we talk about a ton

of other stuff from cybersecurity,

AI scares to just bonding over coffee because we both love coffee.

I'll have that full interview on Network Check Academy. Just check that out.

Anyways, what do we need? Honestly, just a computer. Now, as I mentioned before,

this is Command Line World and this is going to be a Linux-based or Unix-based

system, but we're not leaving out anybody here. You got a Mac,

it works great on Mac. In fact, Daniel Mesler, all he does is use Mac.

He loves it. Linux, of course, if you use Linux is your main desktop,

you'll have a good time. And then Windows, which is what I use WSL,

the Windows subsystem for Linux, so it'll work everywhere.

You have Linux and Linux is everywhere. What a time to be alive, right?

Coffee break. For that coffee break, the setup and install is actually really,

really fast.

I'm going to set up a new machine here in WSL on my Windows machine.

You don't have to do this unless you don't already have one.

And then best practice,

just go ahead and do a pseudo a PT update to update your repos and go ahead and

run a pseudo a PT upgrade if you haven't already. Now on Mac,

you don't have to do that,

just make sure you have your system updates and you'll be golden.

Now let's install Fabric, Mac, windows, Linux.

We're all following along right now.

First we'll just copy and paste this command cloning the Fabric project from Git

Hub. If we type in ls, there's the fabric project right there.

We'll go ahead and jump in there, CD fabric. And then to install everything,

we'll use a tool called PIP x, but we'll have to install PIP X first.

Now on Mac and Windows with WSL, we'll do a pseudo a PT install Pip x.

If you're on Mac, you'll want to use a tool called Brew.

Brew is an amazing utility and I think Package Manager that enables you to

install a ton of things and you should have it on your Mac.

So install brew first if you don't already have it.

And then with the Command Brew, say Brew, install PIP x. Again,

that's Mac only. I'll go ahead and install PIP X. Yep,

and now we'll install Fabric with pip x. Simply type in pip.

I've been sing PIP XA lot, it's making me feel weird. Anyways,

pip x install dot, that should be it. Ready, set,

go and done. We have all these tools installed.

I do have a note from me saying that my path variables aren't correct.

I'll just run this command real quick. You might have to do the same thing.

Pit backs Ensure Path done, and now Fabric is almost ready to go.

We'll just need to run one command fabric, dash dash setup. Oh wait,

I got to refresh my terminal. If you're in Linux,

you'll do source tilda or library sign, whatever you want to call that.

Bash rc. If you're on Mac wll, be Z-S-H-R-C.

If you're using ZSH as your default.

And now we should be able to do fabric dash dash setup.

And what this will do is ask you for a couple things. Your open AI,

API key if you want to use G PT four and all those other models and also your

Anthropic API key to use the cloud models. Now what that means is yes,

you will need an API key. So if you don't already have one, go get one.

I'll put a link down below showing you how to do that.

I'm going to grab mine real quick and I'll paste that there.

And then my Claude API key, which is the Anthropic API key.

And then one more thing, it's going to want your YouTube API key.

So when you are going out to YouTube to pull those transcripts from videos,

you can even do comments. It will use a Google, YouTube, API key.

Those are free to set up. Again,

I'll have a link below to show you how to do that.

And once you've added your API keys, that's it. Now you may be wondering, Chuck,

why do we have to do that? Well, remember,

fabric is just a framework in itself is not ai.

It will use whatever favorite AI you have.

Now that does mean that if you're using Open AI or anthropic,

you're going to have to pay for that usage and it's a pay as you go thing.

Put your credit card in there.

Most of the time it does end up being cheaper than just paying for Chad GPT the

Pro, but just keep that in mind. Now,

if you don't want to pay anything and you don't want to give any data to anybody

ever, there's local LLMs as well, which Fabric just added. Thank you, Daniel.

So if you have Alama installed or llama installed on a remote server like I do

on Terry, we can type in fabric and do dash dash list models.

And right here are the available local models. I'll grab a YouTube video.

This one by Peter McKinnon, I've been meaning to watch.

I'll just grab the summary and then with Fabric,

I'll do a dash dash model to specify a certain model.

In my case it'll be LAMA three colon latest.

Then I'll do dash SP to specify the pattern, which will be Extract Wisdom.

And just like that, I'm using a local model.

Now if I want to use a bigger model like Llama three seven db,

that's not going to run on my local computer, but it will run on Terry.

So to connect to a remote AI server,

specify remote llama server,

put the IP address in of your remote server. This is Terry.

Specify the model Llama three 70 B,

and then your pattern. Now I'm not sure what the whole Alex and Jordan thing,

but that's how you do it. And when I'm away from Terry,

when I'm out of the office remote somewhere, I still want to talk to him.

And here's how I do it. When I'm working remote out and about as I normally do,

I got to make sure I can run my fabric commands and access Terry.

Wherever I am right now, it's not going to work. I'm getting nothing.

That's where Twin Gate comes in.

My favorite way to remotely access my stuff back at home, my office, my studio,

everything. Setting up Twin Gate is pretty stinking easy,

easier than standing in a field. What was that?

All you have to do is set up a free twin gate account, create a network,

and then deploy connector.

It could be a Docker container on a Raspberry Pi in your house or running on

your Sonology nas like I do at my home.

And within a few minutes you get remote access to everything you want to like

now lighting strikes, nah, I'm good. Wildlife maybe. What was that?

But seriously, wherever I go, wherever I am, other hidden holes,

I can remotely access my stuff back at home, including Terry.

My AI server Twin Gate is special because they use all the latest and greatest

technologies to make sure your connection is fast,

including quick one of the new internet protocols that is blazingly fast.

And with Twin Gate, you can control exactly what your people have access to.

All my employees, I don't want you using Terry when you're away from the office,

but you can log into the server and work. I'll allow that.

So if you want to use my remote access solution, check it out, link below.

I've been using it for over a year and it's my favorite way to remotely access

everything. I even did it when I was in Japan too. Worked great. Alright,

I'm getting out of here.

Now let's get a bit more advanced and break down some fabric stuff first.

You don't have to just give it stuff like copy and paste from a YouTube

transcript or something to work with fabric.

You could ask a basic question like watch this. I can echo saying,

give me a list of all ice creams flavors and what year

they originated. Actually, I'm pretty curious about that.

I'll pipe that into fabric and we'll break this down a bit.

So far we've been using the command or the switch.

Sp that's a combined switch. Let's split 'em up so we can talk about it.

So dash s and dash P dash s is for stream. And when we use that switch,

we're telling it to go ahead and output whatever the AI says as it's saying it

stream it to us. PP is for specifying the pattern.

So right after you put the pattern you want to use,

and we'll just say the pattern ai, which is a specific pattern,

just allowing us to talk with AI just told me, no,

you can't do the AI exceeds practical limits. You exceed practical limits.

I'm just kidding. Let's try something more easy. There we go. Now we're talking.

And by the way, when I use fabric about specifying a model,

it defaults to using open AI and GPT-4 Turbo. If you want to change that,

especially if you want to stick to local models,

we can do fabric dash list models to see all our models and then do fabric,

dash, dash change default model and then specify the model.

Now we can also do the command fabric, dash dash list,

just the list and it'll list all the available patterns we have right now.

Again, so many things you can play with. Now I want to show you something crazy.

As Daniel mentioned before, the theme behind fabric is very well, fabric E,

so you got fabric, then you've got patterns. If you want to run a server,

which does some fun stuff, I'm not going to cover right now, it's called a mill.

But you can also do what's called stitching,

which allows you to stitch patterns together. So let's try this.

I've got this article, this long read about that YouTuber poppy,

do you remember her? She's still around. She's crazy. It's super long read.

I'm just going to copy and paste everything and put this into fabric PB paste

fabric. And by the way, I know you're probably wondering, Chuck,

how are you doing that? What is this PB paste thing?

This is built in by default into Max. So if you have a Mac,

just enjoy it on WSL and Linux. It's harder to do.

I'll show you how to do that here in a minute. And if I run out of time,

I'll show you somewhere else. Anyways,

we'll paste that in there and I'll use the prompt, summarize.

So summarize the article,

and then I'll pipe that result into another fabric command or stitch

it. And this pattern will be right essay.

Actually I'll do a dash s so we can see that streamed in and go. Now fabric,

while it's doing this thing, just think about what it's doing right now.

First it's going to summarize that entire article.

Then it's going to kick its summary over to the right essay pattern.

That's powerful. This is crazy. Writing an essay.

We can also do a thing where we analyze the claims of the article.

This is not stitching, I just want to see what happens. Analyze claims,

I mean, this is just cool. Again, these prompts, crowdsourced, open sourced,

they've been meddled with and messed with to make 'em perfect and they're not

perfect. I mean they're still going to be worked on improved.

You can also do one called label and rate giving it a quality score saying

it's B tier, consume when time allows.

This is another superpower of fabric and the idea and the mentality is bringing

two AI and how you might approach your life.

We'll talk more about that here in a minute. I don't want to dive too deep,

but now I want to show you how you can create your own patterns because right

now we're using what's built in default just there.

So I'll show you how you can approach writing a pattern and then getting it into

fabric so you can use it. Keeping in mind that when you write a pattern,

it remains local to you. It doesn't get uploaded to the fabric repository,

and none of that's happening unless you want to submit it. That's up to you.

Everything's still private. But when I first started trying to write patterns,

I didn't really know how to do it.

So I would just go and pick one of my favorite ones, extract wisdom,

and just kind of modify it, which is absolutely a great way to do it.

But then I found this, there's a pattern. There's a pattern for everything. See,

a pattern means solving a problem.

There's a pattern called Improve Prompt that basically does everything for you.

It's crazy. So check this out, we'll do it real quick.

We'll echo something and say you are,

we'll just try to write it on prompt real quick, but messy, dirty.

And by the way, this is a real example of how I wanted to,

and I talked with Daniel about this, how I wanted to digest sermons better.

I go to church every Sunday.

Sometimes I'm serving in the nursery taking care of babies,

and I miss the sermon. Now,

I rarely have time to go back and watch the sermon throughout the week.

So if I could just somehow digest it like this, that'd be amazing.

But I wanted to create a pattern that would look for specific things,

unique to a sermon. So let's try this. Alright,

so I'll pipe that out to fabric and I'll do the pattern and prove prompt.

That's crazy, right? This is so cool. So this is just live off the cuff.

I'm going to take this, copy it.

So now I've got my instruction and I'm going to go to the place where our

patterns live. Here's where they live. We'll go to cd.

We'll do dot config slash fabric type in Ls.

Here we can see we have a directory for patterns.

That's the patterns that fabric we'll use.

And then we have a directory called My Patterns, which is what I created.

So go ahead and make that for yourself right now. We'll do a mic directory,

M-K-D-I-R, let's call it my super Awesome Patterns.

I'll jump in there. And then to create our new pattern,

we'll make a new directory. Call it Sermon Sensei.

It's a has how spell Sensei? Nope. We'll CD in there.

And we'll make a new file called System md. Do nano system md.

Jump in there and I'll paste the contents of that pattern. Control X,

Y enter to safe. So to summarize what we just did,

we created a directory to how our custom patterns and inside that directory,

we made a new directory creating a new pattern sermon sensei,

and we created a file called System md,

which is the system prompt to the actual contents of the pattern. Now,

the reason we created our own super awesome special directory is that often

patterns will be updated because again, this is open sourced,

crowdsourced patterns are always being improved. And if they're in the repo,

you'll want to update your patterns.

So we might do this fabric dash update and that'll update your patterns,

but it will overwrite anything that doesn't belong in there.

So we're keeping our custom patterns inside my super awesome patterns.

That way they're never deleted,

but to make sure they can be used by fabric when we run our commands,

we do need to copy everything into the Patterns folder.

So we'll do that real quick. We'll simply do copy or CP dash r,

we'll specify our directory,

our home directory symbol tilda slash config slash fabric slash your

directory. So mine is super awesome, that super awesome patterns.

We'll do the asterisk to make sure all the folders and stuff are copied just so

into R patterns, directory config, fabric patterns, just like that.

So now if we do fabric dash dash list to list our patterns,

I should see sermon sensei right there.

Notice I have another one called Sermon Wisdom, which I had previously created.

Now let's test out Sermon sensei. I'm going to grab the sermon from my church,

one of our recent ones. Do yt, grab that transcript, put the URL there,

and then pipe that into fabric using my sermon sensei pattern I just

created. Now this is pretty cool and honestly I think it needs some work.

So just like the open source patterns in the fabric repo,

you can work on yours and keep iterating.

So the one I really enjoy is the one I created Sermon Wisdom that I think does a

better job and that really does demonstrate the power of a really,

really good prompt. I mean, I love how it pulls out quotes.

Probably one of my favorite things,

and that was killer is the references pulling out all the scripture or things

that he mentioned in the sermon. I love that. Now looking back at Fabric,

if we type in fabric help,

we can see there are options we haven't mentioned yet.

We're not going to go over all of them,

but one thing I do want to touch on is the idea of a context.

I'll let Daniel talk about that real quick.

This is the latest thing that I've been working on under Config Fabric.

We now have a context file.

My context file is about increasing human flourishing by

helping people identify, articulate, and pursue their purpose in life.

Helping people transition to Human 3.0 to be our best selves.

This is literally my soul that I'm translating it to text.

I haven't done this yet, but I want to make a context for myself soon.

Now I've got two more features.

I want to show you the PV paste option for Linux users and the ability to save

anything you create with Fabric two obsidian, my favorite notes application,

what I use and what I've been obsessing over for a while now. Oh, it's so cool.

I just found this. But before we get there,

I want to talk more about the philosophy behind Fabric and why it's kind of

captured my imagination and why I feel like it's more than just a fancy

prompter. Now, to understand that,

I think we need to know a bit more about Daniel Mesler. Daniel Meer is a hacker.

That's his background. That's what he did and still does.

And the reason I created it is because I basically went independent

as of the end of 22. As soon as the ai,

as soon as GPT-4 launched, I was working at Robinhood at the time.

I built a VM program over there.

And before that I was at Apple and a bunch of other places,

and I had been in AI for five, six years.

But when I saw g BT four happen, I was like, okay, I'm out.

I basically got out and said, I need to do this full time.

So I started collecting all this different AI stuff as I'm sure everyone has

seen. And what I found after a couple of months is like, okay,

I've got a million different prompts. Now what do I actually do with them?

What I started doing is collecting them into an infrastructure that I could use

personally, and this is a little bit before Fabric,

but it became the content for Fabric.

And essentially what it turns into is these patterns here.

And you can tell just by using these patterns that this is the result of many,

many iterations. It does things so well.

And if you've been using AI for a while, again,

you know what it feels like to use a really good prompt and had your results be

so clean and almost exactly what you were looking for or even more than what you

expected. And that's so fun.

Now I want to get back to the idea of human flourishing. Again,

that is the goal of the Fabric Project, and honestly, when you hear that,

it kind of captures your imagination, right? I love, because for me,

I dunno if you're like this, but the more AI advances,

the more I get just a little bit more scared of where my place in the world

might be. But when I hear about projects like Fabric,

where it's not about replacing humans,

but about augmenting humans to help us become better,

to help us think better and to help us consume more content.

And that's one of the main things that Daniel uses this tool for.

One of the main reasons he created this tool is that there's so much content

being produced all the time from YouTube videos to podcasts to articles.

Just staying relevant in your space and your niche takes a tremendous amount of

time. Time. You don't have to consume things.

A big part of it is I am using it to determine what I should go

watch regularly. In fact, I'm actually building a product around that called,

but I won't say what it's called, but I'm building a product around that.

So just because I need it in my life, essentially,

I'm using it as a filter to determine what I should go watch and then go watch

it fully. And oftentimes I take manual notes,

but I watch or listen to it in its entirety,

and then I go and take notes on it and it spawns thoughts.

So I'm not stepping away and disengaging,

I'm still reading massive number of books,

I'm still watching the videos, I'm still reading the essays.

What this is helping me do is just filter out or filter up or

raise attention to the particular stuff that I want to watch.

So a big way he uses fabric is to filter out what is good,

what deserves a long watch or what just needs to be summarized and

quickly digest it.

And it's so cool because the way he created fabric in a lot of these prompts is

he designed these prompts,

these patterns in a way that is meant to mimic the way he would approach

something,

the way he would watch a video and take notes the way he would listen to a

podcast and take notes. It's kind of crazy.

My very first thing that I made for Fabric is the ability to emulate as if I

took slow notes on a piece of content by hand.

And that's what this emulates.

Now a little story time. The past six months,

I've been on a journey of being very particular,

very intentional with what I consume, how I spend my time,

and writing down as many things as I can, taking good notes, processing things.

And that led me to my question I asked Daniel,

and that's if we start using things like fabric,

AI tools to do the processing and thinking for us,

do we lose that value? So the curmudgeon in me, the old man in me,

don't do the AI stuff.

It's going to keep you from becoming a deep thinker and learning how to really

analyze things. But here's what Daniel said about this.

Yeah,

I think the way to use it is to use the

context stuff

that we're starting to build now that's already in the project and basically

define what you're trying to do. You can define inside of the context,

I need to learn this much about these topics,

recommend to me the best courses to do that with.

And then when you take a piece of content and it's overwhelming,

you could put it through fabric and essentially distill it down.

And importantly,

it could tell you what not to distill down because there's so much advantage to

going back to your earlier point,

you don't want to take the weights out of the gym.

So everything shouldn't be a summary.

Sometimes you have to put the hard work in,

but you can use it to tell you or advise you or recommend to you

which things you should do slow and painful and difficult because that's where

you get the most muscle growth.

Don't take the weights out of the gym.

So he'll use fabric to help him determine what should be slowly watched and

processed. We saw that earlier with the label and rate pattern.

It tells you like, oh yeah, you got to watch this right now,

or you could wait on that. Again,

it's not about replacing humans replacing you, it's about making you better,

about taking your current capabilities and using AI to increase that at a

faster rate than you could before.

It's about identifying a problem that you might have and then creating a pattern

to help you solve that problem.

And all the patterns you're seeing there that are built into fabric are a result

of like, I've got this problem. Here's a pattern that can fix it.

And you can do the same thing for your life.

And we talked a bit about a feedback loop to where I may have fabric,

I'll create a pattern for this that I'll look over my journal entries throughout

the week and then it'll tell me,

maybe you didn't read enough that week and that's why you're feeling sad. Or,

Hey, you're feeling fantastic because you ran four times this week.

Keep doing that.

It's that kind of augmentation that I'm really excited about and that's where I

see myself growing the fabric.

And I literally use this all the time for so many things.

Me and Daniel also talked about how we both started recording conversations.

Like any conversation we have with a sibling, a friend, a spouse,

record it, then transcribe it with Whisper ai, which is you can use it locally,

it's free, and then pipe that into fabric.

I actually started doing this recently with, we have a weekly bible study.

We have a core group of people where we meet and we talk about our lives and

things and tell funny stories and talk about what we were learning and going

through.

So I recorded that and I created a pattern that would extract the things I might

care about from those moments. Check this out. It's so cool.

So I have the recording transcribed.

I'll cat that and pipe it into fabric using my GC analyzer.

That's what we call our communities gospel communities.

And I won't show you everything in this because it's very personal,

but I want to show you how cool this is. And honestly,

this whole fabric project is making me rethink about the role of AI in

my life. It's here. Like it or not, what are you going to use it for?

How can you use it to help make you better?

And that was the first video I made ever about ai. Chad GPT came out,

I was terrified, but then after processing it a bit more, I'm like,

you know what? This is going to make us better. It's about us.

It's about humans flourishing. There was a funny moment about,

it was just a funny story about sleepwalking and it found that, that's so cool.

I just did this one. Look at the funny moments here. Yeah,

we talk about weird stuff, but I want to remember that stuff.

I want to be able to go back to that search, have that in my second brain.

That's a video coming soon. And again,

I talked with Daniel about a lot more stuff including coffee.

So if you want to see that entire talk, that entire conversation,

you can check the link below to Network Check Academy, which is my new project,

my baby, where I'm creating it. Courses to help you become awesome in it.

I would love for you to join. We have an amazing community. Go check it out.

Link below. Now, the two more things I wanted to show you First,

this PB paste thing. Actually, you know what,

I'm not going to show you right here. I'm going to make a video about it,

and by the time you watch this video, that one should already be up.

So go here somewhere, it's going to be awesome. Just go and jump there.

But now to the Obsidian Save thing, this is so awesome.

So let's take that same discussion here from my GC

analyzer. I can save that directly to obsidian.

There's a command that is baked into fabric called Save just like this.

And what that will do is save the contents of whatever you're doing in fabric to

a note in obsidian, creating a new note. Now to make that happen,

we first have to tell fabric where our obsidian lives in our operating

system. So if you're new to obsidian,

obsidian is all about just text-based documents. Again, world of text.

All it is just a bunch of markdown files and it's somewhere on my hard drive to

tell fabric where that is. I'll just edit an environment file. So I'll go nano,

jump it to my config folder, fabric folder,

and there should be a file NV for environment variables.

You can see that's also where our API keys are stored.

And here's my path to the directory.

I want to store new mark dime files created by fabric. With that in place,

I want to take away the S option. I don't want to stream it.

I'm just going to save this to a file.

So I'll pipe it out to the save command and then I'll just name it and go.

And then if I jump into my obsidian,

which we're going to shield most of this from you,

if I go to my specific folder, there it is. Auto Magically. That's killer.

Again, it's all about removing friction. This is so amazing. I love this stuff.

Lemme know if you like this too.

Lemme know if you want me to make more videos like this about new AI tools or

just exploring how we can improve ourselves and make ourselves better with the

help of ai. I don't know.