Getting Started

Chat Copilot Sample Application

This sample allows you to build your own integrated large language model (LLM) chat copilot. The sample is built on Microsoft Semantic Kernel and has three components:

1. A frontend React web application
2. A backend .NET REST API service
3. A .NET worker service for processing semantic memory

These quick-start instructions run the sample locally. Extensive documentation can also be found on the official Chat Copilot Microsoft Learn documentation page for getting started.

AI Empower Labs has extended Microsoft's code to support our on-premise Generative AI LLM and RAG containers. Further extensions by AI Empower Labs enable LLM autotooling and function calling with any LLM, not just OpenAI LLMs. These enhancements have been embedded in our extended version of the Copilot.

To deploy the sample to Azure, please view the "Deploying Chat Copilot" section after meeting the requirements described below.

What is in the Package

The Co-pilot is a ready-to-go sample front-end application adapted for AI Empower Labs backend containers. The sample implements a co-pilot application and can be used with AI Empower Labs' innovation on LLM Autotooling to enable function call AI applications and provide a user-friendly interface.

Getting Started with the Public Shared CoPilot Version

You can access the public shared version online and download a copy using the instructions in AI Empower Labs' public open-source GitHub Studio/Co-pilot.

Instructions on how to install the Copilot applications are found there. You can then build your own version, deploy it to a Docker or Kubernetes server, and adjust the appsettings.json files.

The Copilot comes with a couple of sample plugins to illustrate the powerful autotooling capabilities. These can be downloaded separately and activated in the plugin. There are two sample plugins: a web searcher, allowing the LLM to extend itself by searching the internet, and an email sender, a sample API calling application enabling the Copilot to send emails as part of the service it provides.

AI Empower Labs CoPilot Plugins

AI Empower Labs' version of CoPilot comes with some sample plugins. These plugins make use of the LLM autotooling and function calling concept that enables AI to perform actions. We have added a couple of sample plugins in the open-source application to demonstrate the powerful capabilities of our CoPilot.

Sample Plugins

1. Email Sender Plugin: This plugin enables the CoPilot to send emails as part of the service it provides. It showcases how the CoPilot can perform API calls to external services, making it a versatile tool for various applications.

2. Virtual Sommelier Plugin: This is a sample application that enables AI to understand and call an API for a wine shop. The CoPilot can act as a virtual sommelier, understanding what products are available and recommending them from the local store. This plugin demonstrates the possibilities of creating AI workers that are context-aware and company-aware, adhering to company limitations and policies while working.

These sample plugins illustrate the potential of AI Empower Labs' CoPilot to create AI workers that can perform specific tasks within the constraints of a given context, ensuring compliance with company policies and limitations.