This project provides a Python-based creative writing assistant that leverages Azure OpenAI, Bing Grounding, and Azure AI Search to help users generate well-researched articles. It's designed for developers and users who want to build sophisticated AI-powered content creation tools.

How It Works

The solution employs a multi-agent architecture orchestrated by Python. A research agent uses Bing Grounding for topic exploration, a product agent performs semantic similarity searches against a vector store via Azure AI Search, a writer agent synthesizes this information into an article, and an editor agent refines the output. Prompty is used for prompt management and evaluation.

Quick Start & Requirements

• Install/Run: Use azd up after initializing with azd init -t contoso-creative-writer.
• Prerequisites: Azure account with access to Azure OpenAI Service (GPT-4o/GPT-4o-mini, 80TPM recommended), Bing Grounding, and Azure AI Search. Python 3.10+ and Docker Desktop are needed for local development.
• Setup: GitHub Codespaces offers the easiest setup. Local setup involves installing dependencies via pip install -r requirements.txt.
• Links: Prompty Documentation, Multi-agent applications using Azure OpenAI article

Highlighted Details

• Integrates Azure OpenAI, Bing Grounding, and Azure AI Search for comprehensive article generation.
• Utilizes Prompty for efficient prompt engineering and evaluation.
• Supports local testing via FastAPI and includes an optional web interface.
• Offers evaluation metrics for Coherence, Fluency, Relevance, and Groundedness.

Maintenance & Community

This is a Microsoft sample project. For community support, refer to the Microsoft Open Source Code of Conduct.

Licensing & Compatibility

The project is released under the MIT License, allowing for commercial use and integration with closed-source applications.

Limitations & Caveats

The required Azure OpenAI models (GPT-4o, GPT-4o-mini) may not be available in all Azure regions. The README recommends using eastus2.