Devtrails Phase 2: Parametric Insurance Platform with AI-Powered Features

The article discusses the development of GigGuard, a parametric insurance platform for delivery partners in India, and the key AI-powered features implemented in Phase 2 of the project.

Details

The article describes the development of GigGuard, a parametric insurance platform for Zomato and Swiggy delivery partners in India. In Phase 1, the team built a working demo with a Next.js frontend, Node.js backend, PostgreSQL, and a Python ML service for dynamic premium pricing and fraud detection. Phase 2 focused on enhancing the platform's capabilities with various AI-powered features. This included replacing the text-based zone matching with Uber's H3 hexagonal grid for more precise sub-kilometer payout calculations. The team also integrated a Thompson Sampling contextual bandit to learn which coverage tiers are most suitable for different worker segments. Additionally, they ran a SAC reinforcement learning agent in shadow mode to automatically tune premiums against the target loss ratio. The biggest challenge was tackling the problem of GPS spoofing and insurance fraud. The team developed a Behavioral Coherence Score that combines four independent signal layers (accelerometer patterns, battery drain rate, cell tower triangulation, platform online status) to detect illegitimate claims. This approach makes it difficult for spoofers to appear legitimate. The article mentions that the core engine is working, and the team is now exploring further enhancements like GNN-based fraud detection, smart contracts on Polygon, and causal inference for payout validation.