smartFly

Inspiration

Air travel can be stressful, especially when it comes to finding the best flight routes, minimizing layovers, and understanding airport connectivity. While existing travel applications provide search functionalities, they often require users to manually apply filters and interpret complex UIs. We wanted to simplify this process by leveraging advanced AI, graph analytics, and intelligent automation to create a user-friendly, chat-based travel assistant.

What it does

SmartFly is an advanced flight and airport analytics application that utilizes GraphRAG (Graph Retrieval-Augmented Generation) to provide intelligent insights into air travel. It offers:

1) Flight and Airport Queries: Retrieves real-time flight and airport data effortlessly. 2) Shortest Path Analysis: Identifies the most efficient flight routes between destinations. 3) Network Connectivity Analysis: Analyzes airline networks to optimize travel planning. 4) Simplified Chat Interface: Unlike traditional flight search apps, SmartFly allows users to ask questions naturally without dealing with complex menus.

How we built it

1) ArangoDB: Utilized its pre-configured FLIGHTS dataset for graph analytics. 2) GraphRAG Approach: Integrated graph-based retrieval for contextual accuracy. 3) ArangoGraphQAChain: Used to translate natural language queries into AQL (Arango Query Language) and return responses in natural language. 4) NetworkX & Nvidia CuGraph Acceleration: Implemented complex network analysis for travel insights. 5) Hybrid Query Execution: Combined AQL for simple queries and NetworkX for complex queries, merging results using an LLM. 6)ArangoSearch: Leveraged multimodal search capabilities for deeper query understanding.

Challenges we ran into

1) Integrating GraphRAG: Ensuring efficient retrieval of graph-based data while maintaining contextual accuracy. 2) Hybrid Query Execution: Designing a seamless process for handling both simple and complex queries in a unified manner. 3) User Experience: Balancing advanced AI functionalities with a simple and intuitive user interface.

Accomplishments that we're proud of

1) Successfully developed a fully functional AI-powered flight assistant with advanced analytics. Integrated GraphRAG, making SmartFly more accurate and context-aware than traditional flight search tools. 2) Optimized query execution using ArangoDB, NetworkX, and Nvidia acceleration, significantly improving response times. 3) Created a user-friendly chat interface that makes air travel insights more accessible.

What we learned

1) The power of graph-based retrieval in enhancing AI-driven travel insights. 2) How to effectively combine AQL, NetworkX, and LLMs for hybrid query execution. 3) The importance of user-friendly interfaces in AI-driven applications.

What's next for smartFly

1) Real-Time Data Integration: Incorporating live flight updates for real-time recommendations. 2) Multi-Language Support: Expanding accessibility for global users. 3) Provide more than just text, in addition to text the output can also be images.