By Mathew Donfrancessco, Founding Partner, Nuon AI
The heart of Nuon’s Artificial Intelligence software is a set of proprietary algorithms based on Reinforcement Learning (RL). So why do we use RL? In simple terms, RL allows us to run experiments on live quotation and sales data – and learn from them.
In RL-speak this breaks down to:
Observe – look at the quotation data
Action – run an experiment on it
Reward – measure the success.
The RL metaphor
Similar to a robot finding its way around a maze, each random turn builds a picture of the layout, until the robot learns to navigate it. Nuon’s RL feels out the market in the same way, and applies those insights as pricing adjustments to optimise the value of an insurance product.
How to really achieve full market insight
We believe that actively experimenting with live insurance data is the only way to gain full insight into how the market is behaving. Insights obtained passively from historic data can only give an opaque indication of what the market was up to ‘before today’, and worse, any changes that are made to react to those insights are imprecisely measured – for example an uplift in sales could be due to a rating factor change, or conversely could be due to some other market effect such as a competitor raising their prices. Without continuous experimentation there is no real way to be sure.
At Nuon, we turned traditional thinking about the insurance industry on its head. We started by looking at industry-standard problems with fresh eyes, discarding any preconceived ideas about GLMs.
Insurers and pricing models
One thing’s for sure, when it comes to pricing, the insurance industry can do better. When it comes to what the market needs today, AI opens up a whole world of possibilities for insurers, brokers, intermediaries and actuaries.
So we built an AI ecosystem system that does just that. We built it using RL, from the ground up – to offer a new level of pricing sophistication the insurance market needs today.