Project Fabric: An Amazing AI Tool That Augments Human Capabilities And is Changing How we Interact with LLMs to Solve Complex Problems

Humans interact with LLMs based AI using text-based prompts, these prompts are what we use to ask AI to do some tasks. When these prompts are grouped and arranged in a certain order they form a pattern. Patterns can be refined and stitched together to further make the AI model to solve specific problems, in other words, talking to these AI models like they’re humans elicits a better response, better results and this is the whole idea of Fabric.

Fabric is an open-source AI tool designed by Daniel Meer, it utilizes the concept of patterns (a collection of prompts), the patterns are well curated, tuned, refined, stitched together and used to accomplish complex operations. Fabric interacts seamlessly with various AI models, including GPT, Claude, Google (Gemini), and local LLMs (e.g. LLaMA running on your own computer or server) etc. Fabric also allows users to create custom patterns to address specific needs, extending its functionality. More to this, the patterns are open source and crowdsourced which means anyone can contribute to them, making them better over time.

Fabric’s goal is to try and augment humans with AI, to try and reduce that friction, hence fostering human-AI collaboration, and to be able to use AI to solve complex problems. It enables the creation of personalized AI workflows, tailored to individual needs and goals. Fabric empowers individuals to harness AI for personal growth, knowledge acquisition and has the potential to revolutionize information consumption and analysis, hence enhancing productivity. Fabric can be integrated with note-taking applications like Obsidian, facilitating knowledge organization and retrieval.

In Summary, Fabric can be used to:

1. Transcribe and summarize YouTube videos for valuable insights
2. Pattern stitching, allowing users to chain multiple AI tasks for complex operations.
3. Create custom patterns tailored to your specific needs and challenges.
4. Facilitate efficient content consumption by identifying valuable material and summarizing key takeaways.

In this article, we are going to install Fabric on Linux and use it to transcribe YouTube videos and extract wisdom, insights, write an essay, etc.  You will need “sudo” privileges to run the following commands, so at the prompt type “sudo su –” and supply the root password for your system.

Step 1: Update Your OS Packages to the Latest Version

As good practice, it is important to keep your OS packages updated, this helps to maintain the most recent security and bug patches in your system.

Step 2: Clone the Fabric Project from GIT

The fabric project was created and is maintained by Daniel Miessler, and is publicly available on his GitHub Repo.

Change directory to where you want fabric installed and use the “git clone” command

Run the list command to confirm that the fabric directory has been successfully cloned to your local server.

Step 3: Install PIPX

Step 4: Install FABRIC

Install fabric using the PIPX tool that you just installed in step (3) above. To proceed, change to the fabric directory and run “pipx install”

At this point, logout and login again to your CLI to continue the fabric setup process.

Step 5: Setup your Fabric environment

Fabric uses API keys for authentication to the cloud AI models e.g. GPT (OpenAI), Claude or Google (Gemini). At this point you will need to login to the different cloud AI platforms and generate the API keys which you will need to complete the step. You will also need to generate YouTube API key to be able to transcribe YouTube videos.

(i) To create your OpenAI API key and access GPT models using fabric, you need to go to the OpenAI platform. Under API Keys, click on “Create new secret key”

(ii) To create your Google API key and access Gemini models using fabric, you need to go the Google AI Studio.

(iii) To create your YouTube API key and access YouTube data using fabric, you need to go the Google Cloud Console, Navigate to “APIs and services”, and click on “Library”

Scroll down and click on the YouTube Data API, enable the API, go to “CREDENTIALS”, and click on “CREATE CREDENTIALS” to generate API Key.

If you wish to use Claude models with your fabric project, you can head to their Anthropic Console and create an account. However, Claude API and OpenAI API require you to be on a paid plan. Google’s Gemini API is purely free for use up to a certain usage quota. At this point you have your API keys for GPT, Google and YouTube, this is enough to get you started and you are now ready to run the fabric setup command.

Copy and paste the API keys at the respective fabric setup prompt, or press enter to skip the API key that you won’t be using. Fabric setup with download the models and update the patterns and setup the fabric environment.

Step 6: Take your fabric for a spin

Start with listing all the available models in your fabric environment.

Check out all the available fabric patterns

Use the help command to get all the available fabric options

To change the default model, use the command below. It will require you to restart your terminal (re-login) for the default model to take effect. And going forward, all the fabric prompts will use that default model, unless the model is specified in the command with this switch “fabric –model”.

Test the YouTube Transcription and Analysis Function

To update the fabric patterns, use the command below:

One of the most exciting aspects of Fabric is the ability to create your own AI patterns. This allows you to tailor Fabric to your specific needs. You can create solutions for problems unique to you and share them with the community. By training Fabric on your own data, you can further personalize your AI experience. The more you use it, the better it understands your needs. Fabric’s vision goes beyond just making tasks easier. It’s about augmenting our capabilities and enabling us to thrive in an AI-driven world. From summarizing complex research papers to crafting compelling marketing copy, Fabric is transforming workflows across various domains.

I encourage you to explore the potential of Fabric and join the growing community. Together, we can shape the future of AI and use it to empower ourselves and our work. You can run fabric on your windows PC CMD by using the Windows Subsystem for Linux (WSL), a feature that allows developers to run a Linux environment without the need for a separate virtual machine or dual booting.