Crash-risk maps and rider safety scores for two-wheeler delivery fleets
TL;DR Wreckodds reads the data that electric two-wheeler fleets already collect, like harsh braking, speeding and wrong-way riding, and turns it into a map of the junctions and hours where their riders nearly crash. Fleet operators pay a monthly fee per vehicle for safer route hints and weekly risk reports.
Two-wheelers are where India's road deaths pile up. In 2022, 168,491 people died on Indian roads, two-wheeler riders made up 44.5% of them and cyclists another 2.9%[1]. Delivery riders spend more hours on the road than almost anyone, often against the clock, yet I found no public count of how many get hurt.
Fleets feel it as cost. Each crash means a damaged vehicle off the road, a rider off work and, more and more, a welfare question for the platform.
Most fleet telematics in India was built for trucks and buses[2][3], so operators running thousands of delivery scooters get location dots and not much else.
Wreckodds plugs into the IoT feeds that electric scooter fleets already run for tracking and battery management, or into a rider app where there's no device. A model flags near-miss events such as hard braking, sharp swerves and sudden stops, and pins them to the road. Over a few weeks a pattern shows up - a flyover exit where riders brake hard every evening, or a one-way street riders keep taking the wrong way because the legal route adds 900 metres.
Fleet managers get a weekly report by hub and by rider, framed as coaching and not punishment. Riders get short voice alerts before known hotspots and, where a safer route costs under two minutes, a nudge to take it. A camera study of Beijing bike lanes found that lane type, roadside parking and poor surfaces drove daytime risk[4]. Phone and vehicle data should show similar spots without a camera on every vehicle.
I narrowed the seed, which pitched predictive safe routing for every micro-mobility rider. Individual riders rarely pay for safety and free map apps already route them. Fleets pay for downtime, so they're the buyer.
The fleet operator pays - EV rental firms that lease scooters to delivery riders, and in-house fleets at logistics and quick-commerce companies. One shared operator's delivery scooter is used by riders working for the big food-delivery and quick-commerce apps[5]. Truck-fleet telematics firms already sell driver scorecards to Indian fleets[2][3]. Pricing per vehicle per month is my assumption to test. Insurers covering riders are a later buyer.
India's four labour codes took effect on 21 November 2025. Aggregators must now pay 1–2% of annual turnover, capped at 5% of payouts, into social security for gig and platform workers[6]. MoRTH's July 2025 aggregator guidelines require up to ₹10 lakh of health and term insurance for drivers and formalise bike taxis[7]. Rider safety is now a line item on someone's budget.
There's no public figure for spending on two-wheeler fleet safety analytics in India. The narrowest firm number is fleet scale, and the largest shared operator alone runs about 50,000 shared EVs[5].
One telematics firm claims more than 100,000 customers for fleet telematics with driver scoring[2], which shows fleets do buy safety analytics. That's mostly trucks and buses though, so treat it as adjacent.
Video telematics vendors sell systems that flags harsh braking, speeding and phone use, with gamified driver scorecards[2]. Indian truck-telematics firms offer driver behaviour scoring alongside predictive maintenance for trucks and buses[3]. Neither mentions two-wheelers.
Big EV fleet operators hold the raw data but treat it as tracking and battery health. The gap is safety analysis built for scooters in dense city traffic.
| Difficulty | Moderate (41/100) |
| Starting capital | ₹50–95 L |
| Time to launch | 5–8 months |
A small team can build the software. The hard parts are getting fleets to share raw data, telling real near-misses apart from potholes and speed breakers, and proving that the reports actually cut crashes. That proof needs months of before-and-after data from at least one fleet, so it is not a quick win.
No licence is needed to sell analytics software. The weight sits on data and labour rules.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't let a risk score decide pay, fines or firing. The data is noisy and riders will rightly see it as surveillance, so a person reviews any action taken about a rider.
Want to work out the AI layer for Wreckodds? Talk to Yash →
Here's what can go south -
I like this for a founder with data science or telematics experience who can sell to fleet operations heads. If you're expecting consumers to pay for a safety app then don't bother.
Prove first, with one fleet of 500 or more scooters, that the flagged hotspots match real crash and repair records at least 60% of the time.
Plan for roughly ₹50–95 L, mostly on a small data team, integrations with fleet systems and a long pilot.
No licence is needed to sell analytics software. You need DPDP-compliant consent from riders, and you should watch the labour codes if scores ever affect pay.
Yes, delivery fleets face crash risk everywhere. Data protection and gig-work rules differ, so check both before you sign your first fleet.
The free Wreckodds validation kit has the line-by-line capital breakdown, the effort scoring, the data-protection and labour-code checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for fleet operators and riders, and 5 assumptions to test. Download the kit →
Thinking about building Wreckodds or something close to it? The make-or-break work is getting one large fleet to share raw data and prove fewer crashes, and that first fleet deal is something I can help you plan. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.