QBE invests in AI start-up that mines data hidden in documents

This is the insurer's third start-up investment, and aligns with strategy to upend the traditional ways it does business

QBE invests in AI start-up that mines data hidden in documents


By Alicja Grzadkowska

QBE Insurance Group’s streak of investing in technology start-ups continues. Through its investment arm QBE Ventures, the company recently partnered with HyperScience, a machine-learning company putting artificial intelligence to work. The start-up automates office work and will allow QBE to glean useful data from thousands of documents that often just get filed away in a box or the cloud, never to see the light of day again.

“HyperScience, to us, was a really obvious choice and an obvious partner because of the problem they were solving and the way that they were doing it,” said Ted Stuckey, head of QBE’s Global Innovation Lab. “If you look across the normal operations of your standard property and casualty insurer, so much of what we do is back-office processing of information. Just the volume of the documents that we deal with, the volume of human-readable content that is touched by so many different people across the organization, automating that [by] using artificial intelligence to influence and add efficiency to that process was not only a huge operational efficiency game, but an obvious partnership.”

By transforming paper files or PDFs into machine-readable files, HyperScience helps organizations respond to customers faster, facilitate straight-through processing, and access data that’s been hidden in unsearchable documents, according to QBE. For a historically form-heavy industry, the technology gives QBE’s data analytics team an opportunity to mine data and detect fraud or find underwriting and pricing variables, among other useful applications.

“We have a wide variety of different types of documents and forms that we get from our insureds and from other individuals across the insurance value chain that have content that we don’t ever actually structure,” explained Stuckey. “Think of that as damage assessments, engineer reports from underwriting files – things of that nature that have a wealth of content, but because of the way that our systems are structured, or because of the way that things have been built up over time, we don’t actually capture that and we can’t use it for really anything.”

HyperScience tackles an “un-sexy problem,” says Stuckey, and uses advanced AI to find a solution for labour-intensive processes, which made the start-up stand out. QBE’s other investments have also been AI companies – RiskGenius, which analyzes policy and endorsement language and assists in product development, and Cytora, which focuses on improving the underwriting process.

“Our first three investments in many ways had the potential to fundamentally, completely transform our company,” said Stuckey. “We know that we’ve barely cracked the surface on what the potential is for this.”



Keep up with the latest news and events

Join our mailing list, it’s free!