The UK is at the forefront of the machine learning revolution, and the Early Stage investment team here at Amadeus is an active investor in the field. Start-ups applying artificial intelligence (‘AI’) and machine learning to a vast range of products come onto our radar almost daily, but investing in these complex technologies is not for the faint-hearted. So, I thought I would share some insights on start-ups we have backed and ‘where next?’ for investors in a sector offering apparently endless possibilities.

First, a quick analogy to explain what the difference between rule-based and machine learning-based systems is: remember when you were taught it was safe to cross a road when the light was green but not when it was red? That’s just like old-fashioned computer programming: a set of conditions is described with a rule, or an ‘IF this, THEN that’ statement, and some action or operation is chosen. However, as people get older and wiser we come to realise that the colour of the light only tells some of the story, the important information is whether there is any traffic coming or not, and we can rely on experience instead of just the rule. That kind of adaptation through experience is what happens in machine learning, and in the past few years sophisticated mathematical models have been developed which can do this much more accurately than people.

A great example of applying this kind of technology is London-based company Ravelin, whose machine learning technology helps to prevent credit card fraud whilst identifying and responding to the constantly-changing techniques being used by fraudsters. Machine learning can detect such patterns rapidly; Ravelin’s systems respond in a fraction of a second and can operate tirelessly 24/7.

Applying modern artificial intelligence techniques to enable driverless transport has the potential to change not only personal journeys, but also logistics and deliveries. Cambridge- and Bristol-based startup Five AI is using AI to enable vehicles to operate without the need for detailed maps and pictures of all the places the vehicle needs to go, unlike other systems under development. It’s a new approach that could prove critical to the effectiveness of autonomous vehicles, since it allows them to react to live events rather than relying on previously collected and stored mapping data.

Also located in Cambridge, Healx uses machine learning to identify treatments for rare diseases, by sifting through vast databases of genetic and chemical data to identify potential drug candidates. Gene-based medicines are already part of the next wave of healthcare, and using AI will make it possible to identify cost-effective treatments for diseases that affect relatively small groups of people worldwide.

Startup company PROWLER.io has a focus on delivering AI ‘bots’ that behave just like humans in virtual environments. This has obvious applications in gaming but also has implications for other arenas including smart cities, robots and drones. This learning can start in complete safety in a virtual setting before being transferred to real-life contexts.

As these companies show, there are few limits to the application of machine learning. As investors, we find that great technology is a necessary but not a sufficient condition for us to back a business. What these start-ups also have in common is a strong founding team, a vision of how their product will beat its competition and a considered view of its potential market. Finding these diamonds in the rough is what makes so-called ‘early stage’ investing so rewarding.

Alex van Someren is Managing Partner of the Amadeus Early Stage Funds