Announcing the winners of RAGHack 2024!

We loved watching developers build hacks using a variety of techniques. 4 languages ​​(Python, JavaScript, .NET, Java) and Five Azure data stores (Azure Cosmos DB, Azure SQL, Azure Database for PostgreSQL, Azure AI Search). DevRoper made RAG. Hack across a variety of areas including business, sports, travel, wellness, education, society, entertainment, academia and more!

There were so many impressive examples of hacking, but the judges narrowed the list down to 80.+ Up to 10 for our products. Check out the winners below!

Best overall: DocAssistant.Charty

DocAssistant.Charty is amazing. ingestion tool Azure SQL database schema You can answer questions by indexing with Azure AI Search and then using Azure OpenAI to generate SQL statements, charts, and summary answers.

We’ve seen many of our customers request this kind of functionality over the past year, and we’re excited to see an open source solution that demonstrates a user-friendly approach to answering analytics questions about your database.

Technologies used: .NET, Azure AI Search, Cosmos DB, Azure SQL, Azure OpenAI

Best Features of PostgreSQL: Football-Analysis-Copilot

This RAG application demonstrates the power of RAG on a PostgreSQL database using both the azure_ai and azure_local_ai extensions for creating inclusions and the pgVector extension for searching for inclusions across multiple tables. This app can answer your questions about soccer. (known as “football” in the US), based on data collected from external sources.

Judges praised the app for being highly customizable, including multiple options for vector search indexes and Azure OpenAI models to help determine the best combination of parameters.

Technologies used: Python, Azure Database for PostgreSQL Flexible Server, Azure OpenAI

Best of Azure SQL: UniChatbot

Unichatbot Not only can you answer student questions about what courses are available for your major, but you can also register students. Those You can also select courses and select course times that do not overlap. This tool stores all student, course, and enrollment information in an Azure SQL database and then Determine which SQL query to run using: OpenAI features calling. Possible functions are defined using the Pydantic schema in Python backend code and then mapped to corresponding Azure SQL queries. The judges liked this alternative approach of using LLM to interact with database content.

Technologies used: Python, Azure SQL, Azure OpenAI

Best of Cosmos DB: Discord Community Agent

CoMA is a Discord bot that allows people to easily connect with other community members through shared interests and find answers based on past Discord conversations. CoMA stores user data and documents in both Cosmos DB and Azure AI Search, and uses the Azure OpenAI Assistants API to answer questions from Discord users. The judges liked RAG’s integration into a platform where users were already asking questions and its use of technology to bring people together.

Technologies used: TypeScript, Azure AI Search, Azure Cosmos DB, Azure Cosmos DB, Discord.js

Best of Azure AI Search: Manufacturing Support Bots

This chatbot Provide support to manufacturing operators using Azure services. The code includes a custom indexer for Azure AI Search hosted in Azure Functions that ingests data from an Azure SQL database and vectorizes it using Azure OpenAI. The chat application was built using Azure AI Bot Service.

The judges liked the addition of a “rate your answer” step to each question. User feedback is an important and often overlooked aspect of building RAG applications.

see: video | password | submit

Technologies used: Python, AI Search, Azure SQL, Azure AI Bot Service, Azure Functions, Azure App Service

Best Python: StoryWeave

StoryWeave is an interactive text-based game based on vector search. Learn about calling Azure Database for PostgreSQL and OpenAI models. The app first collects a story like “Beauty and the Beast,” divides it into paragraphs, vectorizes the chunks, and stores them in a database. At the same time, we use the GPT model to generate a list of characters and their strengths and weaknesses. The player can then select his character and start interacting with the world by selecting tasks from the generated list. The adventure unfolds with a combination of vector search and GPT calling. The judges liked this method.his app I use RAG in a very different way than most submissions..

Technologies used: Python, Azure Database for PostgreSQL, Azure App Service, Docker, Streamlit, Blob Storage

Best of .NET: ContosoTravelAgency

Contoso Travel Agencyn Advanced travel planning system Using a multi-agent architecture built Uses semantic kernels and Azure durable functions. When a request comes in from a user, ManagerAgent selects which subagent to use, such as FlightAgent or WeatherAgent. Agents can then search the data store, such as FlightAgent, which searches for matching flights in Azure Cosmos DB. These agents are all based on the Azure OpenAI GPT model to select the right action and return user-friendly responses as needed. plusThe Cosmos DB database has been enhanced with native vector embeddings to also handle semantic search. The judges liked the way it combined the functionality of several simple agents into a sophisticated architecture.

Technologies used: .NET, Azure Cosmos DB, Azure Durable Functions, Azure Redis Cache, Semantic Kernel, Blazor, Postmark

Best of Java: Paper Mentor AI

Paper Mentor AI StreamlinedS study process By answering questions and generating summaries based on a vector repository of indexed research papers.. The data collection process uses Azure Document Intelligence to extract text and then uses Spring AI to Vectorize the text and store it in a PostgreSQL database. The backend uses Spring Boot to retrieve the vector store and calls OpenAI, while the frontend uses Next.JS to call the backend. The judges liked that this solution was practical and comprehensive in its implementation.

Technologies used: Java, Spring Boot, Azure Document Intelligence, PostgreSQL, OpenAI

Best JavaScript & Colon; Learning Path Certification Builder

This application helps job seekers identify the skills they lack for the job they want and find the certifications in the Microsoft Learn catalog they need to build those skills. The app uses an Azure SQL database populated with data about Microsoft Learn authentication. Once a user uploads a resume and desired job description, the app uses full-text search to search the database and Azure OpenAI provides suggestions based on matching rows. The judges praised RAG for its original idea and use of RAG for a non-chat-centric user experience.

Technologies used: JavaScript, Azure SQL, Azure Data Studio, Azure OpenAI, NestJS, React, Fluent UI

Best features of Azure AI Studio: MyFitnessBuddy – AI-powered fitness assistant

My Fitness Buddy Apps provided personalized exercise routine And meal planning. The application combines a traditional RAG (using Azure AI Search) and a Graph RAG (getting graph-structured data from Azure Cosmos DB using the Gremlin API). The RAG flow is orchestrated with two endpoints in Azure AI Studio. One is for query rewriting based on intent and the other is for answering questions in a personalized way.

Technologies used: Python, Azure AI Studio, Azure AI Search, Azure Cosmos DBstream light