Azure AI Assistants tool is an experimental Python application and middleware designed to simplify the development, experimentation, testing, and debugging of Assistants created with Azure OpenAI Assistants (Preview) (see below). Use this powerful, easy-to-setup low-code / no code playground tool to quickly experiment and build AI Assistants within your application with Azure OpenAI Assistants API.
Important
The Azure AI Assistant Tool is currently in Alpha. This early stage of development means the project is actively evolving, with significant updates and improvements expected. Users should anticipate changes as we work towards refining features, enhancing functionality, and expanding capabilities. We welcome feedback and contributions during this phase to help shape the future of the tool.
🌟Assistants, a new API from Azure OpenAI Service, is a stateful evolution of the Chat Completions API. Assistants makes it easier for developers to create applications with sophisticated copilot-like experiences in their applications and enable developer access to powerful tools like Code Interpreter and Retrieval. Assistants is built on the same capabilities that power OpenAI’s GPT product and offers unparalleled flexibility for creating a wide range of copilot-like applications. Copilots created with Assistants can sift through data, suggest solutions, and automate tasks and use cases span a wide range: AI-powered product recommender, sales analyst app, coding assistant, employee Q&A chatbot, and more.
Features include:
💬 Inbuilt thread and memory management
📊 Advanced Data Analysis, create data visualizations and solving complex code and math problems with Code Interpreter
🚀 Build your own tools or call external tools and APIs with Function Calling
📚 Retrieval Augmented Generation with File Search tool (coming soon to Azure OpenAI Assistants)
🎤📢 Speech transcription and synthesis using Azure CognitiveServices Speech SDK
📤 Exporting the assistant configuration into simple CLI application
Learn more about Assistants on Azure OpenAI Service:
📹 Watch a short video about Azure OpenAI Assistants 📖 Read the launch announcement 📌 Get familiar with the Assistants API Quickstart
✔️ Enhance Developer Productivity: Streamline the assistant development process with Azure OpenAI Assistans through built-in middleware libraries and tools that making it easy to integrate AI capabilities into your copilot applications
✔️ Enable rapid prototyping: Create amazing demos with AOAI Assistants and develop end-to-end assistant solutions with a robust set of features, including built-in system functions, dynamic generation of user functions specification and implementation, assistant task creation and scheduling, and much more.
✔️Optimize your copilot development workflow: Get a reliable and scalable framework to test new Copilot use cases and dynamic AI applications with Assistants API without the need to build out manual tooling and configurations
- Easy Configuration: Set up your assistant with the model, custom instructions, files, and tools
- Tool Integration: Incorporate knowledge retrieval, code interpreters, and built-in system and dynamic user functions to enhance assistant skills and capabilities.
- Dynamic User Functions: Quickly create and apply user-defined functions to assistants.
- Task Management: Efficiently manage and schedule tasks, including batch and multi-step operations, for parallel execution.
- Create an Azure Subscription for free, if you don't have one already
- Apply for access to Azure OpenAI Service in this Azure Subscription. Azure OpenAI Service is currently a limited access service so access is granted through an application process. Most applications are processed within a day of applying.
- Azure OpenAI Assistants is currently available in Sweden Central, East US 2, and Australia East. We are expanding our models and regional availability - see the models guide for the latest model and regional availability for Assistants.
- Create an Azure OpenAI resource on Azure Portal with one of the Assistants supported models from the models guide page deployed in that region
The Azure AI Assistant tool requires Python >= 3.8 on your system. You can download Python from the official website. Latest stable Python version is recommended.
Create a new Python virtual environment. Virtual environment is recommended for safe install the SDK packages:
- On MacOS and Linux run:
python3 -m venv .venv source .venv/bin/activate - On Windows run:
py -3 -m venv .venv .venv\scripts\activate
Run the following command in your terminal to install all the necessary dependencies as specified in the requirements.txt file.
pip install -r requirements.txt
We have included the azure.ai.assistant library currently within this repository as part of the alpha status of the project.
The plan is to release library on PyPI for more convenience installation in future.
Build the wheel for azure.ai.assistant library using the following instructions, or use the wheel package provided under the repo's release tags directly.
- Python wheel and setuptools packages are required to build the library package. You can install them using the commands:
pip install wheelandpip install setuptools - Go to the
sdk/azure-ai-assistantfolder - Build the wheel using following command:
python setup.py sdist bdist_wheel - Go to generated
distfolder - Install the generated wheel using following command:
pip install --force-reinstall azure_ai_assistant-0.x.xa1-py3-none-any.whl- This installation will pick the necessary dependencies for the library (openai, python-Levenshtein, fuzzywuzzy, Pillow, requests)
To successfully make a call against the Azure OpenAI service, you'll need the following:
- ENDPOINT: This value can be found in the Keys and Endpoint section when examining your resource from the Azure portal. Alternatively, you can find the value in Azure OpenAI Studio > Playground > View code. An example endpoint is: https://docs-test-001.openai.azure.com/.
- API-KEY: This value can be found in the Keys and Endpoint section when examining your resource from the Azure portal. You can use either KEY1 or KEY2.
- MODEL DEPLOYMENT-NAME: This value will correspond to the custom name you chose for your deployment when you deployed a model. This value can be found under Resource Management > Model Deployments in the Azure portal or alternatively under Management > Deployments in Azure OpenAI Studio.
Next, go to your resource in the Azure portal. The Keys and Endpoint can be found in the Resource Management section (see image below). Copy your endpoint and access key as you'll need both for authenticating your API calls. You can use either KEY1 or KEY2. Always having two keys allows you to securely rotate and regenerate keys without causing a service disruption.
Create and assign persistent environment variables for your key and endpoint.
- Set the Azure OpenAI Service key, endpoint. Version is optional and default currently is
2024-05-01-previewfor assistants.
Windows: Use setx or set command depending on your preference.
setx AZURE_OPENAI_API_KEY "Your Azure OpenAI Key"
setx AZURE_OPENAI_ENDPOINT "Your OpenAI Endpoint"
setx AZURE_OPENAI_API_VERSION "Azure OpenAI version"
setx OPENAI_API_KEY "Your OpenAI Key"
Linux/Mac
export AZURE_OPENAI_API_KEY="Your Azure OpenAI Key"
export AZURE_OPENAI_ENDPOINT="Your OpenAI Endpoint"
export AZURE_OPENAI_API_VERSION="Azure OpenAI version"
export OPENAI_API_KEY="Your OpenAI Key"
- Set Cognitive Services Speech key (this is optional and if you want to use speech input & output).
Windows:
setx AZURE_AI_SPEECH_KEY "Your Speech Key"
setx AZURE_AI_SPEECH_REGION "Your Speech Region"
Linux/Mac
export AZURE_AI_SPEECH_KEY="Your Speech Key"
export AZURE_AI_SPEECH_REGION="Your Speech Region"
In the root of this repository, command:
python main.py
This command will start the Azure AI Assistant Tool and you can interact with it through its user interface which looks something like this:
The Azure AI Assistant Tool is licensed under the MIT License. See the LICENSE file for more details.
The Azure AI Assistant Tool uses PySide6, which is licensed under the GNU Lesser General Public License (LGPL). By using PySide6, you are able to modify and redistribute the library under the same license. For more information on PySide6's license, please visit Qt Licensing.
We welcome contributions and suggestions! Please see the contributing guidelines for details.
This project has adopted the Microsoft Open Source Code of Conduct. Please see the code of conduct for details.
If you find a bug in the source code or a mistake in the documentation, feel free to submit bug report. Or even better you could submit a pull request with a fix.
If there's an feature that you'd like to see added, feel free to file a Feature Request.
If you'd like to implement it yourself, please refer to our contributing guidelines.