RouteShare

Inspiration

Carpooling can make daily travel more affordable and sustainable, but finding a suitable ride often requires users to manually search through routes, compare pickup points, and keep track of trip updates. We wanted to make this experience more natural by allowing users to interact with RouteShare through voice. Instead of navigating through multiple screens, drivers and passengers can simply ask an AI assistant to find trips, check booking information, or get updates about their journey.

What We Built

RouteShare is an AI-assisted, stop-based carpooling platform that matches drivers and passengers based on overlapping routes and recommends convenient pickup and drop-off points. For the Alexa+ track, we are building a voice-first assistant that acts as a conversational interface to RouteShare. Users can interact naturally with the system to:

• Find suitable carpooling trips
• Check trip and booking details
• Get pickup and drop-off information
• Ask about upcoming trips
• Receive useful updates during their journey The assistant connects conversational requests to RouteShare capabilities through an MCP-based integration, allowing the AI to interact with the application's services instead of simply returning static information. ## How We Built It The core RouteShare platform is built around a NestJS modular monolith with an event-driven architecture. The system uses GraphQL for application APIs and PostgreSQL/PostGIS for route and geospatial data. The Alexa+ experience is designed as an additional conversational layer on top of RouteShare. The assistant interprets natural-language requests, determines the appropriate action, and communicates with RouteShare through our MCP integration. This architecture allows the voice experience to remain decoupled from the core application while still accessing real application data and functionality. ## Challenges One of the main challenges was deciding which functionality should be exposed through the conversational interface and how to make complex carpooling operations understandable through natural language. Another challenge was designing the interaction between the AI assistant and the existing RouteShare backend while keeping authentication, data access, and business logic inside the application's backend. We also had to think carefully about how voice interactions should handle ambiguous requests, such as multiple matching trips or pickup locations. ## What We Learned Building RouteShare for the Alexa+ track helped us think beyond traditional mobile interfaces and explore how conversational AI can become another interface for an existing application. We learned how to design application capabilities specifically for AI agents, how MCP can connect an agent with real application functionality, and how voice interactions require a different approach to UX compared with traditional screen-based interfaces. Our goal is to make carpooling feel less like searching through an application and more like simply asking an assistant to help.

Built With

• amazon-web-services
• event-driven
• graphql
• mobile
• node.js
• voice