58 Claims Lloyd’s Lab Impact Report 59 Collaborative intelligence “In this case, our job was to tackle a huge for faster claims processing challenge and to take a step forward for the whole market. Claims automation has been a Cohort 3, 2019 goal for many years, but insurers needed a market-wide programme to innovate in this area; which is, of course, what the Lab is designed to facilitate. In this case, everyone Some challenges in our market require bravery even to approach, and the knew what we were trying to achieve was automation of parts of the claims process is one such example. The difficult and we collaborated to work at it bene昀椀ts are huge – if AI can be used to analyse claims documents together. This approach, along with the accurately, it becomes possible to approve the payment of claims to technology, absolutely delivered, and is a customers much faster. It’s an area where innovation could transform great demonstration of what can be customer outcomes. However, the challenges are also big. achieved in an open innovation space.” The data pool for London market claims is very deep, and there are sometimes dozens of linked policy Ed Gaze, Lloyd’s Lab manager documents that must be analysed – previously always by a human claims handler – to understand the technicalities of a single claim. As a result, the automation of P&C claims checking was once thought to be impossible, and few 昀椀rms were willing to even attempt it. But through the innovation focus of the Lab, the AI 昀椀rm Tautona was enabled to make the impossible real. Before entering the Lab, Tautona had already developed an early stage claims adjustment tool, designed to read each claim and its linked policy documents to produce a recommendation on whether a claim should be paid. The product was conceptually right, and already in use in other industries, as well as in The claims tool that was subsequently rolled out consumer insurance. The challenge was to bring it into has been trained on hundreds of thousands of the P&C insurance market successfully. claims and reads them as a human would, but During their time in the Lab, Tautona was matched with more accurately and within seconds, instead of Atrium and they agreed to work on a pilot to automate 2-3 hours. Tautona says that while human beings analyse claims at 95-96% accuracy, their system “ Having been in the Lab is a door opener for will deliver 98% accuracy. us. We are able to say who we have worked For Tautona, the Lab was a fantastic learning with and that brings so much trust straight experience and sharpened their focus. Previously, the team was focused on end-to-end claims away because they know we understand validation. However, they found a ‘golden nugget’ in their world.” the Lab – P&C policy coverage analysis – where automation could deliver huge bene昀椀ts in a big Tautona market. The Lab also opened up the US market for Tautona, and the 昀椀rm now works with carriers and TPAs in Canada and the US. Atrium’s delegated authority property claims for the US market. During the ten-week programme, they were together able to create a proof-of-concept model which showed how using the right AI could reduce costs and processing times while maintaining accuracy of results. © Lloyd’s