News · Mobility & Logistics

PathPulse AI uses smartphones for road safety

By Startup Enthusiast · 17 January 2026

What they do

PathPulse AI is an AI-first platform that uses smartphones and computer vision to improve road safety

It has raised $900,000 to date and runs pilots in three countries including South Africa

The company plans to launch an on-device Vision Language Model for road understanding in 2-3 months

PathPulse AI is an AI-first platform that uses low-cost smartphones and computer vision to improve road safety and urban mobility.
The company has raised $900,000 to date from undisclosed investors.
Its app 'Scout' collects road data from drivers and has over 50,000 users with 4,500 daily active users.
Drivers from Zepto, Swiggy, Zomato, Ola, and Uber earn money by passively capturing road data.
The platform has distributed ₹12 lakh to drivers across 25 countries.
Pilots are running in three countries including South Africa.
The company is working on an on-device Vision Language Model (VLM) for contextual road understanding, expected to launch in 2-3 months.
Capital is allocated to product development, engineering hires, marketing, and user retention.

The idea came from observing that traffic signals and road infrastructure operate on static configurations without real-time intelligence.
Running complex AI models on inexpensive smartphones is a challenge, so the company built lightweight efficient models.
Selling to governments involves long sales cycles, bureaucratic hurdles, and trust deficits.

PathPulse AI is an AI-first road infrastructure intelligence platform.
It uses low-cost smartphones, computer vision, and AI for road safety and urban mobility.
The app 'Scout' collects road data via drivers, with over 50,000 users and 4,500 daily active users.
Drivers from Zepto, Swiggy, Zomato, Ola, and Uber earn by passively capturing road data.
Data use cases include traffic signal optimization, urban planning, navigation, geospatial analytics, law enforcement, insurance risk assessment, and autonomous vehicle training.
The platform has proprietary object detection and segmentation for potholes, traffic signals, lane markings, and helmet violations.
Pilots are running in three countries including South Africa.

Working on an on-device Vision Language Model (VLM) for contextual road understanding, launch expected in 2-3 months.
Exploring night vision and lidar integration with Middle Eastern partners.
Capital allocated to product development, engineering hires, marketing, and user retention.