If artificial intelligence can learn to drive safely on Indian roads, it can probably work anywhere in the world. That belief sits at the core of drivebuddyAI, an Ahmedabad-based mobility technology startup building AI-powered driver monitoring and advanced driver assistance systems for fleets and vehicle manufacturers. 

Founded in 2018 by electronics engineer Nisarg Pandya, the company is turning the complexity of Indian traffic into a training ground for algorithms designed to reduce accidents today and enable autonomous driving tomorrow.

There are easier places to train machines to understand traffic. Road environments in Europe and North America are far more predictable. Lane markings are clearer, vehicle behaviour is more disciplined, and the number of unexpected objects on the road is limited. 

India offers the opposite. A highway can suddenly merge into a busy village street. Auto-rickshaws, two-wheelers, trucks, buses and pedestrians often move in unpredictable patterns. Lane discipline is fluid. And every city has its own traffic personality.

For drivebuddyAI, this complexity is not a problem to avoid but an advantage to learn from. “We realised very early that a lot of AI models trained on foreign data simply did not work properly on Indian roads,” says Nisarg Pandya, founder and CEO of drivebuddyAI. “So the first thing we did was start collecting our own data.”

That decision defined the company’s approach. Rather than adapting imported models, drivebuddyAI began building its own dataset and technology stack from the ground up.

What began as an aftermarket AI dashcam solution for fleets has gradually evolved into a broader mobility intelligence platform spanning advanced driver assistance systems (ADAS), driver monitoring systems (DMS), fleet analytics and insurance-linked risk scoring.

Today, the company is pitching its technology not only to fleet operators but also to vehicle manufacturers, Tier-1 suppliers and insurance companies, and is preparing to take its products to global markets.

From an Engineer’s Idea to a Mobility Startup

The Ahmedabad-based company began in 2018 as an aftermarket AI dash cam business for fleets. Its founder, Nisarg Pandya, was not chasing hype. He was trying to solve a very real problem. 

Trucks, buses and commercial vehicles were running for long hours. Drivers were getting tired, distracted, and sometimes dangerously careless. The tools available then were basic. Telematics could track the vehicle. A conventional dash cam could record what had happened. But neither could intervene in the moment. Pandya wanted to build a system that could.

“We started as an aftermarket system,” he says. “The idea was to build our own hardware, our own AI, our own software, everything. We built everything from scratch.”

That decision shaped the company’s DNA. DrivebuddyAI did not buy an off-the-shelf device, relabel it and sell it as an Indian product. It built prototypes, mounted them on vehicles, sent them onto roads, and learned the hard way.

That mattered because India is not one road environment. It is many. A Mumbai-Pune truck route is different from intra-city transit in Pune. A tourist vehicle running inter-city has different scenarios from a gas carrier, a staff transport vehicle or a state bus. And then there are the uniquely Indian objects on the road: auto-rickshaws, two-wheelers, cattle, unmarked turns, broken lane discipline and vehicles moving in ways global datasets never anticipated.

In Pandya’s telling, the first breakthrough was not figuring out what worked. It was figuring out what did not.

Pandya is an electronics engineer from Gujarat. He graduated in 2012 in electronics and communication and worked on vehicle tracking systems, GPS hardware and software before turning entrepreneur. He also holds a master’s degree in VLSI and embedded system design.  

That background shows in the way he talks. He does not romanticise the startup journey. He describes it like a debugging exercise. “2016 was the time when I got this idea,” he says. “Let me build something like this.”

The idea took shape in 2018. The first challenge was AI. Today, that sounds almost ordinary. But in 2018, India-specific driving data for computer vision was thin. So drivebuddyAI started with available foreign datasets, trained its early models, then discovered the obvious problem: they did not map cleanly onto Indian roads.

So the team began collecting its own data. They built six or seven prototypes. They put them into different categories of vehicles. They gathered outside-road video first. Then customers asked for cabin-side visibility too. So the company began combining external and internal monitoring. One camera watched the road. Another watched the driver. Eventually, these became integrated systems.

The early years were lean. Even now, Pandya says the company has around 25 people, with roughly 15 in engineering. Much of the core work, he says, was done by three or four people over several years.

That small-team discipline is still visible in the company’s strategy. It likes modularity. It likes reusability. It likes building two steps ahead.

The Funding Stage

Pandya says hardware was a hard sell to Indian venture investors in 2018. The burn profile was higher. The timelines were longer. The appetite was limited. 

In 2019, drivebuddyAI found a backer in Roadzen, an insurtech and mobility technology company that later listed on Nasdaq. Roadzen now owns 75% of drivebuddyAI, while the founding team holds 25%, according to Pandya.

The timing was dramatic. Pandya says the deal closed just before Covid, and the first meaningful chunk of funding came in March 2020, when the world was entering lockdown. Since then, he says, drivebuddyAI has raised around $5 million from Roadzen.

That relationship is more than capital. It also explains the company’s wider ambition. In India, ADAS, telematics and insurance have often evolved in parallel. In the US and some other markets, they are more tightly linked. Safer fleets can mean lower claims. Better driver data can mean smarter underwriting. Video and event data can improve first notice of loss and claims handling.

That bridge between safety and insurance is one reason drivebuddyAI has always looked beyond the camera itself.

What the Product Does

On the surface, drivebuddyAI sells a familiar promise: make fleets safer. Underneath that is a layered operating model. The system watches both the road and the cabin. If a driver is using a phone, not wearing a seatbelt, showing signs of fatigue, or drifting into a risky state, the device gives a real-time alert. The voice prompt can be delivered in regional languages. The aim is to make the device feel less like surveillance and more like an in-cabin assistant.

“We don’t want to fight with the driver,” Pandya says. “Our approach is more human. The system will talk to you. The system will give you voice command. You listen to the voice commands and react.”

That matters. In many fleets, the first reaction to in-cabin cameras is suspicion. Drivers assume they are being watched. Pandya says that changes over time. Once they realise the system is there for safety and coaching, acceptance improves. Some fleets then build operating protocols on top of the alerts. If a fatigue warning repeats multiple times, the driver is expected to stop. Fleet control rooms monitor higher-risk vehicles. 

The economics can be sold in simple fleet language. Better driver behaviour also lowers asset damage and improves operational efficiency. Fewer accidents mean less downtime. Less downtime means more revenue days on the road. 

Roadzen’s public communications have added some external numbers to this pitch. In late 2025, the company said drivebuddyAI had been trained on over 3.5 billion kilometres of real-world driving data and that fleet deployments had shown reductions in drowsiness and overall fleet risk. 

Fleets First, Then OEMs

DrivebuddyAI’s first scale came from fleets and aftermarket customers. It now works both with direct fleet operators and with enterprises that mandate safety systems for third-party transporters.

Pandya gives the example of Supergas, part of Netherlands-based SHV Energy, which transports LPG in India through a network of plants and transporters. DrivebuddyAI has also worked with customers in chemicals, petroleum, cement and logistics.

Then came a second leg: OEM-linked aftermarket integrations.

With Volvo Eicher, drivebuddyAI says it has integrated with the MyEicher platform so customers can see safety data alongside GPS and fuel information in a single system. Pandya says the company is also working with CV OEMs to integrate its systems into their fleet solutions.

The third leg is the hardest and the most important: direct OEM and Tier-1 integration for compliance and factory fitment.

Here, drivebuddyAI’s pitch is modular. If an OEM has a hardware partner but needs software, it can supply software only. If the OEM needs a full stack, the company can provide both hardware and software. It says it is already working with some EV OEMs and Tier-1 suppliers at different levels of integration.

That modularity is central to its survival strategy. It is a small company entering markets where giants operate.

Pandya names competitors carefully. In fleet safety, Netradyne is a frequent rival in large bids. In driver monitoring, he points to Emotion3D among the relevant players. Mobileye, he says, is not really a direct competitor in the same sense. It is an established global peer with a very different position in the OEM market.

A Company that Wants to do More

DrivebuddyAI is not trying to remain confined to one product category.

Its current offerings already span dual-camera ADAS and driver monitoring systems (DMS), six-camera solutions with blind-spot and payload visibility, and modular offerings. Large fleet customers typically pay a monthly subscription that includes hardware, software, connectivity, analytics and after-sales service.

But founder Nisarg Pandya believes the real opportunity lies beyond the device itself. His ambition is to turn drivebuddyAI into a multi-layer artificial intelligence platform for mobility. “We are building AI that solves multiple problems at once,” he says.

Pandya outlines three long-term intelligence layers the company is working toward. The first is fleet inspection. Here, the system analyses driver behaviour to identify risks, improve safety and help fleet managers understand how individual drivers perform on the road.

The second is insurance assessment. By analysing long-term driving behaviour, the platform can help insurers assess risk more accurately and potentially support usage-based insurance models.

The third is vehicle autonomy. By studying how human drivers react to real-world scenarios, the system aims to build an intelligence layer that could eventually assist, and in some situations even control, vehicle behaviour.

Pandya is clear that autonomy remains a long-term goal. But the roadmap is already evolving. In 2026, drivebuddyAI introduced radar integration alongside its camera systems, enabling radar-camera fusion. Until recently, the platform relied primarily on computer vision.

The next step is moving beyond warnings toward limited intervention capabilities, such as emergency braking assistance.

The underlying principle is straightforward. The more kilometres the system observes, the more edge cases it encounters, and the better its decision-making becomes.

The Global Expansion

For most startups, global expansion is a slogan. For drivebuddyAI, it has started to look more operational.

The company made its first appearance at CES 2026 in Las Vegas, where Roadzen presented drivebuddyAI as a vision-first AI mobility platform for fleets, OEMs, insurers and smart-city planning. Roadzen’s CES announcement said the platform was trained on more than 3.5 billion kilometres of data and validated under Indian and European standards.  

Pandya says the company is now expanding into the US, Australia and South Africa, using Roadzen’s existing relationships in insurance and fleet ecosystems as an entry wedge. That is a practical advantage many Indian mobility startups do not have.

He is also looking at adjacent data businesses. Mapping companies, for example, can use non-video insights derived from fleet movement to identify truck parking patterns, points of interest, and road usage behaviour. That opens the door to a revenue stream built not on hardware or subscriptions, but on structured mobility intelligence.

By FY2027, Pandya says, the company is targeting around $5 million in revenue. By 2030, with global OEM programs and adjacent data businesses in place, he believes the business could potentially cross $50 million. Those are ambitious targets. The more immediate milestone is profitability. He says break-even should come once deployments cross 10,000 to 15,000 vehicles, which he expects within about 18 months.

The Next Question

DrivebuddyAI does not have to prove that fleet safety is a real need. That part is done. Its real challenge is scale.

Can a 25-person company move fast enough as global players crowd into India? Can it convert fleet learning into OEM-grade durability? Can it keep funding long-cycle R&D while building a healthy current business? Can it become more than a feature vendor?

Pandya’s answer, at least for now, is to do what the company has done since the beginning: stay lean, build modularly, and treat India’s complexity as an asset.

In a sector that often sells clean futurism, drivebuddyAI’s story is messier and perhaps more believable. It starts with tired truck drivers, difficult roads, and cameras that had to learn local chaos. It moves through insurance, coaching, patents and compliance. And it ends, at least for now, with a company that wants to turn Indian driving disorder into a globally useful intelligence stack.

That is a bold claim. But if the future of mobility is going to be built from data, then the roads that confuse machines the most may also teach them the most.