During cyclones, people often don't know which escape routes will remain passable until it's too late.
Create a routing engine that ingests weather models, terrain data, and infrastructure maps to predict which roads will flood or collapse first, updating in real-time as the storm progresses.
Local governments and emergency services would pay for premium access to distribute to citizens, while individuals in high-risk zones might subscribe for personal alerts.
Start with a simple overlay on existing map APIs showing predicted cutoff times for major roads in one high-risk region.
The biggest risk is inaccurate predictions leading to dangerous route suggestions during actual emergencies.