Special Features
Innovative AI-driven Travel Solutions
to Elevate Your Experience
Short Demo
Experience the Future of Travel
Watch this demo to explore the powerful AI-driven capabilities and architecture behind ContosoTravelAgency. We’ll also share insights into the challenges we overcame during development.
Building the Semantic Layer
Consolidating Complex Data
ContosoTravelAgency integrates a robust Semantic Layer that simplifies querying across complex relationships in multiple tables like Airlines, Flights, Bookings, and Passengers. The Semantic Layer aggregates data into a unified, intuitive view, enhancing the accuracy of LLM-generated insights.
Scaling with Azure Durable Functions
Robust Multi-Agent System
We implemented the entire multi-AI agent system using Azure Durable Functions, which allows individual agents to scale seamlessly. Durable Functions are perfect for handling durable tasks and orchestrating complex workflows. By leveraging patterns like fan-out and fan-in, we efficiently retrieve results from both Natural Language to SQL queries and Natural Language to Embeddings. This enables us to fetch relevant documents with vector search, combining them for a more accurate final result.
Leveraging Cosmos DB Change Feed
Real-time Data Updates
Using Azure Cosmos DB’s Change Feed, ContosoTravelAgency ensures real-time updates to the Semantic Layer. Changes to bookings, passengers, or flights are instantly reflected, allowing for up-to-date insights and responses from the AI agents.
Optimizing with Vector Search
Efficient Retrieval with Vectorization
ContosoTravelAgency enhances its Semantic Layer with data vectorization for faster, more relevant document retrieval. This is particularly useful for queries based on the semantics of the data, allowing for deeper insights and improved user experiences.
