At AI Everything, the buzz wasn’t just about artificial intelligence—it was about how AI is reshaping the way we move. Autonomous vehicles stole the spotlight, from sleek self-driving taxis to AI-powered shuttle buses. 

But tucked in the corner of SteerAI’s booth was something different: a rugged, fully autonomous buggy, built for more than just smooth city streets. It was a clear statement that autonomy isn’t just about urban transport—it’s about tackling the toughest environments, from logistics hubs to military zones.

To dive deeper into this evolving landscape, The Byteline sat down with Michael Sonderby, Acting CEO of SteerAI. As an industry expert, Sonderby was eager to discuss what sets SteerAI apart from other self-driving solutions and the company’s work on Level 4 autonomy—far beyond the assisted-driving tech found in most vehicles today. 

He shared insights on the company’s partnership with A2RL, the challenges of deploying AV technology at scale, and how AI is transforming everything from defense operations to fleet management.

In this exclusive interview, Sonderby unpacks the present and future of autonomous mobility—from high-performance AI-driven race cars to self-driving logistics fleets. He also reveals the biggest hurdles in bringing autonomy to the mainstream and why the industry’s focus should go beyond just replacing human drivers. Here’s what he had to say.

How does your autonomous technology differentiate itself from other self-driving solutions currently in the market?

First of all, there are different levels of autonomy—from zero, which is essentially your cruise control, to five, which is fully hands-off with nobody driving. We operate at level four, the next highest, which basically means it's not the usual L2++ that you will find in many cars these days.

We also operate within very specific settings, meaning it's an industrial one within logistics. It can be ports, warehouses, and we also focus a lot on defense. These two environments basically mean that we need to be able to operate fully autonomously at level four.

In defense, it means unmapped environments and very difficult terrain, so obstacle detection is something that is very important. That's why we have AI powering our software.

I think those would be the main differences—the fact that we can leverage our technology in those very different environments. I think it's also important to highlight what we don’t do because being specific about the capabilities and limitations of these technologies is crucial.

We're not in the passenger or urban environment because that's an entirely different setting. It comes with a lot of complexity, and while safety is always important, it is even more critical in passenger and urban environments.

Can you explain how SteerAI kits power A2RL vehicles and what advantages they bring to autonomous transport?

So essentially, an AV kit is a piece of technology that includes sensors and software, which you install on a vehicle equipped with drive-by-wire. This means we can fit it to almost any vehicle within the context I just shared with you.

With A2RL, we’ve built a partnership where they provide the vehicle chassis—whether it's a race car or a buggy—and make it drive-by-wire. We then install an AV kit on top, allowing different teams to program their software to perform under the various conditions required by the race. In this setup, we act as a technology provider to A2RL, supporting all the teams and enabling these competitive races.

What are the biggest challenges in deploying autonomous vehicles at scale, and how is your company addressing them?

The first challenge is ensuring that autonomous platforms solve real problems with tangible impact. We often engage in discussions where innovation and sustainability are important, but at the end of the day, businesses need to operate profitably. Understanding the core business opportunity is essential for making informed and financially viable decisions in this space.

The second challenge is the speed of deployment—how quickly we can move from proof of concept (PoC) to industrialization. This is particularly relevant to AV kit providers, as we need sufficient time to ensure our solutions function correctly in their intended environments, address key problems, and deliver the expected safety and value.

There is also a time lag in development, but we anticipate that AI will gradually enhance our engineering capabilities. While we don’t see immediate results, AI is expected to streamline the development cycle over time, allowing us to work faster and more cost-effectively. Ultimately, the challenge lies in both helping customers understand the value of autonomous solutions and improving our ability to deliver better, faster, and more affordable technology.

How does your technology handle complex urban environments and unpredictable road conditions compared to traditional self-driving systems?

This is where the testing and learning of AI becomes paramount, and that’s why it takes such an incredibly long time. As you begin testing in a desert environment, the algorithms start to learn about different circumstances, including specific geographical characteristics. For example, sand behaves differently in various parts of the desert—some areas have more stable terrain, while others are more fluid.

Beyond that, AI also relies on additional data points to inform its decision-making. It considers factors such as how fast the wheels are moving through the sand and integrates weather information—analyzing wind speeds and other environmental elements that may affect the terrain and vehicle traction.

Although it doesn’t rain frequently, when it does, it can significantly impact overall performance. Testing is crucial, but equally important is incorporating extra data sources. In defense and security applications, this sometimes includes satellite imaging to ensure vehicles can operate effectively in their intended environments.

What role does AI play in ensuring safety, decision-making, and efficiency in your autonomous vehicle ecosystem?

I believe that any technology has some margin of error. The important point is that these errors occur far less frequently than with human drivers. All available statistics indicate that even with near-perfect autonomous technologies today, our roads would be significantly safer if deployed at scale compared to human-driven vehicles.

I drive almost every day between Dubai and Abu Dhabi, and it's not particularly safe—there aren’t many autonomous vehicles on the road. That’s something important to keep in mind. The key is ensuring that these technologies are thoroughly tested before deploying them on a large scale in urban areas.

We are seeing this process unfold—Waymo, for example, started in a specific area in the U.S. and has since expanded significantly. In the UAE, WeRide has also begun working with Uber. The more we can validate these systems through rigorous testing, the better.

Another crucial aspect is safeguarding against bad actors. Autonomous vehicles rely on software, and it's well-documented that even aircraft systems have been hacked because they aren’t entirely hardware-dependent. The same vulnerability exists for vehicles.

There are two key layers to consider: first, the technical ability of autonomous systems to recognize and respond accurately in traffic, and second, ensuring that there is no external interference with these systems. Protecting vehicles from potential cyber threats is just as critical as their ability to navigate safely.

When do you think autonomous vehicle technology should be sold at large scale, and when should it be made available to consumers?

I think the best place to start with these technology-heavy and cost-intensive solutions is in the public domain. Buses are a good example because they operate in environments with various obstacles, but their fixed routes make implementation easier. Unlike autonomous cars, which need to navigate different roads and optimize routes dynamically, buses follow predefined paths, making them a logical starting point.

The next question is when to make the transition to large-scale deployment. For many transportation companies, this is a strategic decision. Do they need to replace their existing fleet, or can they retrofit autonomous kits onto older vehicles? Deploying AV technology on outdated buses may not make sense. However, if a company is already considering electric vehicles (EVs), then adding autonomous capabilities becomes a natural extension.

On the other hand, even newer combustion engine vehicles could benefit from some level of autonomy. Implementing AV technology can increase efficiency, optimize asset utilization, and reduce dependence on human drivers, potentially improving service quality.

For example, WeRide taxis have been tested in China, and the driving experience has generally been better than human-driven counterparts. While monetization remains a question, there is a clear service-level benefit to AV adoption.

Ultimately, the key factors influencing AV deployment are the age of the fleet and the operator's strategic priorities—particularly in areas like sustainability and the transition to EVs, which naturally complement autonomous technology.

What are Steer.AI's future plans and how do you see the future of the autonomous industry evolving?

I think we are a growing business, and winning the exhibit demonstrated the strong demand for these types of solutions, as well as the curiosity from both customers and potential users. We are evolving by ensuring that we focus our scarce, high-quality talent and resources on a few key verticals.

One of our main focus areas is defense, where we have both the capability and expertise to offer differentiated solutions to customers in defense and law enforcement. Additionally, we are working on the logistics sector, building our own technology as an AV solution provider with the goal of becoming the leading player in the region.

We are also exploring partnerships across various industries, including e-commerce, where there is significant interest. While we may eventually deploy our own solutions, we are currently looking at integrating with the right partners who have complementary technologies.

Strategically, we aim to collaborate with different players in the AV space to expand our suite of offerings. 

This approach helps us scale effectively. A critical aspect of our software is fleet management systems, which are essential for companies considering autonomous technologies. It's not just about having one autonomous vehicle to solve a problem—businesses need an entire fleet, and a robust fleet management system is key to integrating these solutions into their existing operations.