rohit-lakhanpal/build-your-own-copilot

Resources and examples for creating Co-Pilot applications that use modern AI to assist with complex tasks.

Build your own Co-Pilot | Home

powered by

[ Introduction ] → Getting Started → Use Cases → Resources

Table of Contents

1. About The Project
2. What is a Co-Pilot?
3. Technologies Explored
4. The Gen AI Buzz
   • Enabling Everyone
5. Contributing
6. Disclaimer
7. License
8. Contact
9. Acknowledgments
10. References

About The Project

It's fascinating coming across this article titled "Democratizing AI" from 2016 by Microsoft, where the seeds were already being sown for a world where AI technologies are accessible and beneficial to everyone, not a luxury confined to the domains of developers and data scientists. Even back then, Microsoft said:

"It's not about having AI that beats humans in games; it's about helping everyone achieve more — humans and machines working together to make the world a better place."

Fast-forward to today, and we witness a monumental push to integrate AI into every possible application, leading to the emergence of the "Co-Pilot" concept.

This project contains resources and examples to help you as you build your own Co-Pilot applications that use modern AI to assist with complex tasks.

(back to top)

What is a Co-Pilot?

"The concept of a Co-Pilot is an application that uses modern AI to assist with complex, cognitive tasks." - Kevin Scott

Through this Build Your Own Copilot project, we've collected resources aimed at enabling you to Explore, Envision & GTM with your own Co-Pilot applications.

(back to top)

Technologies Explored

These usecases leverage the following technologies:

• Azure AI Speech (learn more)
• Azure AI Vision (learn more)
• Azure AI Language (learn more)
• Azure AI Content Safety (learn more)
• Azure OpenAI (learn more)
• Azure Machine Learning (learn more)
• Azure Cognitive Search (learn more)

Major cloud technologies/frameworks/libraries are listed here:

• 
• 
• 
• 
• 

(back to top)

The Gen AI Buzz

The LLMS/Foundation models have brought about a paradigm shift in AI capabilities, particularly in two key areas: Completions and Embeddings.

The Completions capability of LLMS/Foundation models has revolutionized the way we interact with AI. This feature allows the model to generate creative and contextually relevant text based on a given prompt. It’s like having a conversation with the model, where you provide an input and the model generates a meaningful continuation. This capability is not just limited to conversations, but extends to generating stories, code, essays, and more. The models have been trained on a diverse range of internet text, enabling them to handle a wide array of topics with ease.

On the other hand, “Vector Search” and “Embeddings” are powerful capabilities. Instead of merely matching keywords in a query, these technologies permit the model to grasp the semantic meaning behind a given input, thereby facilitating more accurate and contextually relevant responses.

(back to top)

Enabling Everyone

The advent of Generation AI is nothing short of transformational, representing a paradigm shift in the integration and application of AI across various domains. It heralds a future where artificial intelligence is not merely a tool but a collaborative partner, capable of reshaping industries, fostering innovation, and catalyzing unprecedented growth and development. Gen AI embodies the culmination of advancements in machine learning, natural language processing, and cognitive computing, collectively ushering in an era of limitless possibilities and solutions.

For businesses to truly leverage the full spectrum of opportunities offered by Gen AI, it is crucial to cultivate an environment where everyone is empowered and enabled to contribute. This includes the hackers, who can push the boundaries of what is technically possible; the hipsters, who bring design thinking and creativity to forge user-centric solutions; and the hustlers, who drive the vision and execution to bring innovations to market. By embracing diversity and fostering a collaborative ecosystem that values the unique skills and perspectives of each individual, businesses can unlock the true potential of Gen AI and drive forward the frontier of what is possible in this exciting new era.

(back to top)

Contributing

Contributions are what make the open source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.

If you have a suggestion that would make this better, please fork the repo and create a pull request. You can also simply open an issue with the tag "enhancement". Don't forget to give the project a star! Thanks again!

1. Fork the Project: Fork the repo you want to contribute to by clicking the Fork button on the top right corner of the repo page.
2. Clone the Repo: Clone the forked repo to your local machine using the command (git clone URL_OF_FORK).
3. Branch: Create your Feature Branch (git checkout -b feature/AmazingFeature)
4. Commit: Commit your Changes (git commit -m 'Add some AmazingFeature')
5. Push to the Branch: (git push origin feature/AmazingFeature)
6. Open a Pull Request: Go to your forked repo on GitHub and click Contribute and then Open a pull request. Fill out the details of your pull request and submit it.

Learn more about contributing to projects here.

(back to top)

Disclaimer

This template is provided "as is" without warranty of any kind, whether express or implied. Use at your own risk! The author will not be liable for any losses or damages associated with the use of this template.

It is intended to be used as a starting point for your own project and not as a final product.

(back to top)

License

Distributed under the MIT License. See LICENSE.txt for more information.

(back to top)

Contact

• Michael Friedrich | @github
• Rohit Lakhanpal | @github | [email protected]

Project Link: https://github.com/rohit-lakhanpal/build-your-own-copilot

(back to top)

Acknowledgments

These toolkits are never the result of solitary efforts. I wish to extend my heartfelt thanks to my friends, colleagues, and fellow community members for their exceptional contributions. We have built upon your work, and it is your efforts that have laid the foundation for our success. Your work is not only recognised but deeply valued.

• Prompt engineering samples from my friend and colleague Daniel Scott-Raynsford
• The Azure OpenAI Hack Pack built by my friend and colleague Abby Shen
• Project Miyagi
• Semantic Kernel
• Project Autogen
• This amazing ReadME template

(back to top)

References

• AI Hackathon Starter Kit
• Extend Microsoft 365 Copilot via Plugins
• Teams AI Library
• Models
• Quotas and limits
• Data privacy with Azure OpenAI
• Azure Samples - E2E

(back to top)

Documentation generated by AI, edited by humans. ❤️

(back to top)

build-your-own-copilot's Issues

The Deploy to Azure buttons don't work.

Create a publish the azuredeploy arm templates to deploy cloud resources.

Config centralised for convenience.

Relocate the “config” to a central location so that it’s accessible to all notebooks.

'completion' to 'completions'

config/settings.json.sample line 14:
// Specifies the type of model to use; either "completion" or "embeddings". Only required when ApiType is "azure".

Please replace 'completion' to 'completions':
// Specifies the type of model to use; either "completions" or "embeddings". Only required when ApiType is "azure".

Missing Azure ML

GETTING-STARTED.md needs a section around pre-requisites to create your first prompt flow (e.g. Creating an ML workspace)

Fix breaking changes introduced in Microsoft.SemanticKernel version 1.0.0-beta2

The breaking changes introduced in Microsoft.SemanticKernel version 1.0.0-beta2 are as follows:

• The term "Skills" has been replaced with "Plugins" to avoid confusion³.
• Memory Plugin extraction has been done in two parts¹.
• The Microsoft.SemanticKernel.TemplateEngine.PromptTemplateEngine package has been renamed¹.
• Models collection has been added to PromptTemplateConfig to support multiple AI models¹.
• SKParameterAttribute and related code have been removed¹.
• Semantic functions have been extracted from Kernel.Core¹.
• ImportXXXPlugins methods have been renamed to ImportXXXFunctions for GRPC¹.
• PlannerConfig classes have been refactored for better organization¹.
• Updates have been made to SKContext.KernelContext Phase 2¹.
• IServiceConfig has been removed as it is not used¹.
• IsSemantic and IsAsync properties have been removed from ISKFunction and FunctionView¹.
• FunctionsView class has been replaced with List¹.
• New result types - FunctionResult and KernelResult have been introduced¹.
• ContextVariables have been made into a proper Dictionary¹.

Please note that these changes are part of the ongoing efforts to improve the Semantic Kernel SDK and may require adjustments in your code if you are using this library³. For more details, you can check the official release notes on GitHub.

Source: Conversation with Bing, 23/10/2023
(1) Introducing the v1.0.0 Beta1 for the .NET Semantic Kernel SDK. https://devblogs.microsoft.com/semantic-kernel/introducing-the-v1-0-0-beta1-for-the-net-semantic-kernel-sdk/.
(2) Releases · microsoft/semantic-kernel · GitHub. https://github.com/microsoft/semantic-kernel/releases.
(3) Releases · microsoft/semantic-kernel · GitHub. https://github.com/microsoft/semantic-kernel/releases.
(4) Package Microsoft.SemanticKernel · GitHub. https://github.com/orgs/microsoft/packages/nuget/package/Microsoft.SemanticKernel.
(5) NuGet Gallery | Microsoft.SemanticKernel 1.0.0-beta2. https://www.nuget.org/packages/Microsoft.SemanticKernel/.