At the recent Unbound event in London, I interviewed seven startup founders to find out more about the highs and lows of creating a successful tech business in the UK. Here’s the interview with Vishal Chatrath, CEO of Prowler.io.
ES: So we’re here to talk about Prowler.io. Tell us about this platform.
VC: We are the world’s first principled AI decision-making platform. What this means is we use machine learning to understand, guide and optimise millions of micro-decisions in complex systems, specifically complex and dynamic systems like massively multiplayer online games, smart city simulations, and the key word here is “principled”. What we are doing here is using interpretable principles of mathematics for learning and decision making.
You’ve got some really interesting applications of it in spaces such as smart cities and driverless cars as well, is that right?
Yes, smart cities and driverless cars are the applications that we are working on. They have a very common element about them, that the environment in which they’re operating is uncertain and dynamic by its very nature. You have to have a probabilistic model of that environment and also it consists of millions of agents going around so you have to have the ability to do decision making in a principled way, and that means it has to be highly interpretable.
You don’t want a situation where all these cars and robotic arms are moving around in a city and you don’t know why these decisions are being made. What you don’t want is a black box, and that’s typically a problem with deep neural nets, so that’s something that we don’t do. We focus on bringing the three fields of probabilistic modelling, principled learning and game theory together. We’re the first company to do that.
You’re a relatively young company having been founded at the beginning of last year, and you managed to raise funding in September of the same year. Why do you think you were able to raise that capital in such a short space of time?
I think one of the benefits we had was that the founding team wasn’t doing this for the first time. We had been entrepreneurs before, we’d worked with startups before and, even better, we’d been in the area of machine learning for some time. When we looked at the area of machine learning, almost every other company out there was focusing on deep neural nets to solve perception problems and various recognition problems. You know like a dog is a dog, a cat is a cat, speech recognition and so on. So that we thought was a fairly solved problem from a business perspective and there wasn’t anything new out there.
We soon realised that nobody was using machine learning for principled decision making, and to do that wasn’t a trivial problem. We realised that you have to bring three branches of science together which, until we started, were three completely different areas. Again: probabilistic modelling, principal learning and game theory. They were so distinct that they had separate conferences, separate parties, separate pubs. It is interesting that we have very experienced people in the area of probabilistic modelling, who’d been in the area of academia and research for 20 years – they hadn’t met anyone from game theory in their entire life until they joined our company, so that really explains the disparity.
Of course when we were bringing all these people together with the potential to solve the big problem, that attracted the investors. Hence we were able to grow the team very quickly. We are 30 plus and over 20 are PhD’s and we are based in Cambridge.
What was the reason for setting up in Cambridge, why not London?
I don’t believe that the people live in cities; they live with them. There’s a very clear distinction. When you live with a city it’s a function of a very symbiotic relationship where you absorb the strengths, the beauty and the knowledge of that city. Cambridge is filled with machine learning researchers and professors, so it wasn’t a bad place to start. For the past few hundred years, Cambridge has been the place where the fundamental principles of mathematics that we use for our daily lives were started. So you absolutely couldn’t have a better place to start. As an entrepreneur and a CEO, the ideas that are seeping through the walls at Cambridge, it’s like being a kid in a candy shop.
Have there been any challenges from being based in Cambridge? I know that here in London there’s a real startup and support network, have you ever felt you were lacking that in any way in Cambridge?
No we haven’t. I don’t want to sound ageist here but with the kind of work we are doing we do attract a certain segment of people who are usually quite experienced – usually with 10-25 years experience in their various fields. They are at a stage in their lives where they do prefer a slightly more peaceful environment, which Cambridge does offer, some have families and Cambridge has fantastic schools. I read a very interesting stat somewhere – I’m not sure how true it is but I’ll say it anyway – that 43% of the population of Cambridge has a PhD, so you can’t have a better place to build a machine learning startup.
What about graduates, would you say there’s a pool coming from the academic institutions that are based there or do you lose a lot of those to elsewhere?
We do get a lot of fresh talent also. The way we’ve balanced the team and that I would advise anybody ever starting a startup, is first get your co-leadership team in place. By leadership team I mean really experienced people who know how to build teams and have different functions like engineering, quality etc. Once we had the core team in place then we started to get fresh talent from Cambridge University.
One point I’d like to say is that we’ve hired people from all over the world. The first thing I did when we had the vision to start this company was literally go around the world and find the top people in each of these fields and convince them to move to Cambridge. So it’s not just Cambridge people. We have people from every continent apart from Africa yet, and I’m looking forward to hiring my first employee from Africa.
Talent aside, what are the challenges you’ve faced in creating the company?
The biggest challenge we face is the sheer amount of confusion around the field of AI. That ranges on one end from hype to the other end of the spectrum which is paranoia. One of the biggest challenges has been to explain to people the hype created by deep neural nets. Somehow it seems to drive the whole agenda – people seem to confuse AI with deep learning. Deep learning is just one very small corner of the entire spectrum of machine learning, and of course deep learning is fantastic for a lot of applications and it has shown us some benefits. But we strongly believe that the benefits you’ve seen of machine learning due to deep learning so far is just a trickle. The flood is still to come and that flood will come when you start to see the benefits of using principled AI for decision making.
What would be your main piece of advice for those setting up companies?
I’d say three things. First, aim high and aim best. There are lots of problems to solve in this world and machine learning is starting to make it possible. The field of these problems is extreme so go out and be the best at what you do. Second, when you hire your partners or employees, look for people who are better than you in everything that needs to be done. If they’re not better than you, you might as well do that job yourself. Your job as a CEO is to create the right atmosphere for all these people to work together. Lastly, enjoy and have fun.