Overview
- Gaurav Sen published a demonstration in mid‑August 2026 that used a dashcam with GPS and an accelerometer to detect potholes, estimate their size, and produce structured complaint records during a single commute.
- The prototype sends images and sensor data to AI vision models and developer tools Sen said he built with Codex and OpenAI to classify defects and geolocate them while driving.
- Sen’s demo also said the system searches roughly 2,900 digitised road contracts to match each defect to a tender, contractor and responsible officer, producing complaints with photo, coordinates and tender details.
- The video went viral on social media and drew praise and pilot proposals for other regions such as Shillong, but no Bengaluru civic body has confirmed formal adoption or independent verification of accuracy.
- Reporters and experts flag open issues for real‑world use: detection error rates and false positives, access to complete contract data, privacy and workflow integration with existing complaint portals, and the need for engineering verification before enforcement.