SyncPath: Autonomous Last-Mile Transit Orchestrator

The Problem

Commuters often face "fragmented journeys." A 5-minute delay in a Chennai Metro train or local EMU can lead to missed Uber/Auto bookings, increased cancellations, or long waits in high-demand zones. Current maps are static; they tell you where to go, but they don't act when things go wrong.

The Solution: SyncPath

SyncPath is an agentic AI framework built to unify fragmented transit data. It doesn't just monitor; it takes autonomous corrective actions to ensure a seamless commute.

Key Features

• The Watcher Agent: Polls live GTFS (General Transit Feed Specification) data to track the user’s specific vehicle in real-time.

• The Logistics Agent: Monitors ride-sharing availability and surge pricing at the upcoming destination station.

• Autonomous Re-routing: If a delay is detected, the agents collaborate to automatically suggest a later pickup or switch modes (e.g., from Auto to Bike-Taxi) to maintain the user's schedule.

• Privacy-First & Open Source: Built using local LLM support (Ollama) so user location data isn't unnecessarily stored in the cloud.

Technical Implementation

• Orchestration: CrewAI / LangGraph for managing agent handovers.

• Backend: FastAPI for high-performance asynchronous API handling.

• Intelligence: GPT-4o-mini (for complex reasoning) or Llama 3 via Ollama (for privacy).