Large Language: GPT4 Can Handle Insurance But Isn’t Ready for Core Processes Just Yet (w/Tom Wilde, CEO Indico)

Tom Wilde is CEO at Indico, an intelligent intake solution automating critical middle office workflows across insurance, banking, real estate and healthcare.

'Intelligent intake' at Indico means the starting point for insurers' most critical workflows including underwriting, claims and policy servicing. The common denominator is data culled from emails or other documents, taken from an unstructured format to feed core platforms and decision systems. 

The original digital transformation started with the new customer-facing mobile device, but transitioned through the back office then into the middle office of insurance where business-critical decisions take place. Tom describes actuaries as the original data scientists, taking data and turning them into decisions.  Their modern engines were designed to run on clean sets of data, but unfortunately the reality is that most data lives as unstructured and requires manual rework before it can be used.

Indico was early on understanding the impact of deep learning and large language models, and were the first to build a large language model that allows customers to customize their own machine learning models and solve unstructured data challenges. 

Tom feels strongly that generative AI (eg. GPT4) can easily handle the context of insurance, because the depth of context-understanding enabled by trillions of parameters can understand any context a human can. There are however three main problems to commercially scaling insurance use cases today using GPT4: it is extremely computationally expensive, insurers' data is still sensitive and proprietary, and GPT4 essentially makes up an answer (lies) when it doesn't know one.  Until these large language problems get solved, insurers will be held back from using generative AI in core processes. 

Prior to joining Indico, Tom was the Chief Product Officer at Cxense, a data management provider and founder of Ramp, an enterprise video content management company. Tom worked at Fast Search, Miva Systems, and Lycos. He has an MBA in Entrepreneurial Management from Wharton.

Follow the Insurtech Leadership Podcast airing weekly hosted by Joshua R. Hollander. We give you up-close access and personal insights from the leaders of the fastest-growing #insurtechs and most innovative #insurance carriers and brokers.