Skip to content

Welcome to Meet GPT! This allows business users to experience Azure Open AI Models within their own environment using a PowerApps client.

License

Notifications You must be signed in to change notification settings

pauldatta/meetGPT

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

21 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Meet GPT

Welcome to Meet GPT! This allows business users to experience Azure Open AI Models within their own environment using a PowerApp client.

Features

  • Generates user input responses using Azure Open AI service API
  • Stores user data in Dataverse tables for easy analysis
  • Admin model driven app to manage user access and rate limits
  • More features to come!

Screenshot of Meet GPT app Screenshot of Admin GPT app

Pre-requisites

Before using Meet GPT, please ensure you have the following:

  • An Azure Open AI service account with a deployed GPT-3.5 model.
  • The OpenAI Key and OpenAI URL for your Azure Open AI service deployment.

If you haven't set up an Azure Open AI service account and deployed the GPT-3.5 model, follow these steps:

  1. Sign in to the Azure portal at https://portal.azure.com.
  2. Create a new Azure Open AI resource. Learn how to create an Azure Open AI resource with this Microsoft Learn guide.
  3. Follow the instructions to deploy the GPT-3.5 model within your Azure Open AI service. Refer to the Microsoft Learn guide for detailed steps on deploying the GPT-3.5 model.

Retrieve the OpenAI Key and OpenAI URL from your Azure Open AI service account.

Usage

To use Meet GPT, follow these steps:

  1. Deploy the app to your PowerPlatform environment by importing the PowerPlatform solution in the ./latest folder. You can learn how to import a PowerPlatform solution by following the steps in this Microsoft Learn guide.
  2. Set the OpenAI_Key and OpenAI_URL environment variables to the appropriate values for your Azure Open AI service deployment. Use the entire OpenAI URL including the deployment name and API version Screenshot of OpenAI Studio Code Sample
  3. The mgpt_getResponse flow communicates with the Azure Open AI service deployment API.

Changelog

  • Version 1.0.0.7:
    • New Feature - Favorites, added a flow and UI
    • UI changes, improved layout and design
    • Added tabs for current session, history and favorites, Prompt search
  • Version 1.0.0.6:
    • Added an admin model driven app, still WIP but you can change rate limits
    • Added soft rate limits where
      • Value of -1 : Unlimited API calls
      • Value of 0 : Cant make any API calls
      • Value > 0: Number of API calls, so a user with Rate Limit set to 100 can press send a 100 times
    • Updated the Flow to track limits and also calculate tokens sent
    • Updated the UI to show the rate limit with no enforcement
    • New users get added to the User's table automatically with a default Rate Limit of 100.
  • Version 1.0.0.5:
    • UI changes, improved layout and design
    • Fixed GalleryChat message loading filters, clear and submit action datasource refresh
  • Version 1.0.0.4:
    • UI changes, removed unused screens
    • Only list interactions from today, clear now resets the gallery
    • Canvas app is now version controlled, check the canvas branch
  • Version 1.0.0.3:
    • Fixing the missing/blank app issue.
  • Version 1.0.0.2:
    • Removed an unused HTTP with Azure AD connector from the solution.
  • Version 1.0.0.1:
    • This app lets you send prompts to Azure Open AI and adjust temperature settings through a user-friendly interface.
    • Created in a fast and fun hackathon mode in just 4 hours.

Known Bugs:

  • Gallery view and UI design are being improved.
  • Message send/chat flow is being improved.
  • Has not been tested enough to be production ready.

Roadmap

  • [✓] Admin section to manage user access and token/rate limits
  • [✓] User Favorites with search
  • [WIP] Workflow for users to request more tokens with simple gamification
  • [Planned] Prompt analysis, for admin to monitor user activity and feedback
  • [Planned] Submit an idea section for business users to share AI use cases for their work
  • [Planned] Advanced mode for users to send custom prompts and export JSON etc like Azure Open AI Studio
  • [Planned] Add a datasource (experimental, might start with SPO/OneDrive) and Q&A

Contributing

If you would like to contribute to Meet GPT, please follow these guidelines:

  1. Fork the repository and create a new branch for your changes.
  2. Make your changes and test them thoroughly.
  3. Export your PowerPlatform solution as a zip file.
  4. Submit a pull request with a detailed description of your changes and why they are needed, including the exported PowerPlatform solution zip file.
  5. Wait for feedback and address any issues that are raised.

Microsoft Open Source Code of Conduct

This project has adopted the Microsoft Open Source Code of Conduct.

Resources:

Contact

If you have any questions or encounter any issues with Meet GPT, please create an issue.

Acknowledgements

Meet GPT utilizes the following technologies and resources:

  • PowerPlatform
  • Azure Open AI Service

Special thanks -

  • David E for his the idea and feedback.
  • The Merakist for UX/mockups, design.

About

Welcome to Meet GPT! This allows business users to experience Azure Open AI Models within their own environment using a PowerApps client.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published