Alex Kendall

Wayve is the pioneering start-up developing  Embodied AI for self-driving technology on a global scale.

The Balderton team first met Alex Kendall in his Cambridge dorm room in 2015 to talk about his recently published paper on monocular depth perception and other novel ideas related to computer vision. His PhD work focused on the idea of a robot that could see for itself and understand what is happening – and this fuelled his conviction in the future of intelligent robots and embodied AI. Wayve was born from this thesis, and Balderton led the company’s Series A in 2019.

Wayve is now building AV2.0—a next-generation autonomous driving system that can quickly and safely adapt to new driving domains anywhere in the world. James Wise sat down with Alex to discuss his founder journey and why we will first experience embodied AI through driving. Selected excerpts are below and you can watch the full conversation here.

First, what is AI’s promise and, more specifically, embodied AI?

The great promise of AI is that it can transform the things we do and allow us to do so much more with our lives. If we think about AI today, most people associate it with Chat-GPT, Co-pilot, or other chatbots – something you interact with through your keyboard. Embodied AI offers a physical interface for AI technology to assist us and help us achieve more in our lives more safely and sustainably. Self-driving cars are the best example of this. They will likely be the first chance people get to experience an embodied AI – one that can deliver goods and transport people more safely and sustainably than the transportation we have today. But there’s so much more embodied AI can unlock – whether in manufacturing, domestic robotics or delivery robots.

Driving is a heavily regulated area where safety is paramount and an industry with immense competition. Why start with self-driving cars?

The opportunity is massive. We will see the first large-scale example of embodied AI on the road because there is a clear business case for it, and we also have access to data to solve it. Tens of millions of vehicles produced each year collect data and allow us to train systems that can understand the world. The business case is clear: there is enormous value to society in moving people and goods in a more sustainable and effective manner.

What was your Day 1 vision for Wayve, and how has that evolved?

Throughout my whole life, I’ve been an engineer at heart – I was always interested in the physical world and how we interact with it, from a childhood in the outdoors in New Zealand to some of the technology I played with growing up. My fascination with embodied AI is rooted in the combination of these factors.

I was excited and passionate about a future where we could trust intelligent robots from Day 1, but the path we’ve been on since has meandered and transformed along the way. It is clear to me today that it’s important not to jump suddenly to full-on automation but to transition to assistants and accelerate to full automation. Getting the system out there and giving people exposure to it as a driver assistance system in the broader sense not only provides the system with a diversity of experiences to learn to be safe but also helps people build trust in the technology. A gradual pathway that allows us to develop the best product experience and for people to learn how to interact with these systems is the best path forward.

Going back to your previous experiences, you have a PhD in deep learning and computer vision. Is that necessary for founders tackling these complex challenges like embodied AI? 

I certainly don’t think a PhD is necessary. I was fortunate with my PhD experience because I had a very creative platform to connect with amazing people and the space to innovate and explore commercial opportunities. It was great training to be a deep tech entrepreneur, but there are many other ways to get those skills. Having the fundamental passion and drive for the problem space is vital, especially when building something of this magnitude. It has been six years since we founded the company, but I started thinking about these ideas many years earlier.

How have you personally found this journey with Wayve?

I’ve loved it. It has been an absolute privilege to have this journey. If I think about where we started, where it’s gone, and where we are going, it’s just magical. The challenges just have really changed year on year, which I’ve enjoyed. In the early days of Wayve, when we worked out of a residential house in Cambridge and built our first autonomous car in the garage, we got reinforcement learning to actually learn to drive this car with no human rules or correction when it started to drive off the road. It learned how to lane follow by itself, which has been such a memorable moment. We put a video up on YouTube and it went viral because at the time, in 2018, reinforcement learning had only really been shown to work on video games. So to see positive results on a real robot was an incredible breakthrough for us.

And then of course, the challenges evolved from writing the first code, to building a team, to finding a commercial strategy or leading an executive team now – they’re all totally different skill sets, which I feel fortunate to have had the opportunity to grow into. There have been things you learn the hard way as you go, but having the support of the team around me for that has been amazing.