Case study: how Atrium incorporated Tautona's cognitive claims automation tech

The process by which Atrium Underwriters Ltd. Incorporated Tautona's technology took about three years

Case study: how Atrium incorporated Tautona's cognitive claims automation tech


By Mark Hollmer

Jonathon Hopper (pictured) first became impressed with Tautona during Lloyd’s of London’s Lloyd’s Lab innovation program, which focuses on accelerating the development and creation of new products and services that help its customers. That interest eventually involved into an ongoing partnership.

Hopper is claims manager for Atrium Underwriters Ltd. (Atrium), the managing agent for Lloyd’s Syndicate 609. Tautona is a South Africa-based cognitive automation company whose presentation during the Lloyd’s Lab program drew Hopper’s immediate notice. He offered to be a mentor and the relationship soon blossomed into much more.

“When we were looking at claims areas, in particular with Tautona, they had some really interesting technology, and they came in looking to see how they might be able to apply that,” Hopper said. “They had some South African products that were being developed and used at the time, but they were branching out and looking to London, and that’s how we met them and when we started working together.”

Tautona’s website bills itself as an “intelligence or hyper automation company” that uses AI and robotics to automate insurance industry tasks that traditionally needed “human cognitive abilities.” It focuses exclusively on claim processing for insurers, with its “library of reusable automated claim processes.” The pitch is that the technology can help bring down operational costs, reduce claim handling times and boost the customer experience.

Atrium is using Tautona’s artificial intelligence to automate the initial evaluation of its North American property claims.

Tautona CEO John Holdsworth expressed confidence in the relationship when it was announced in late January.

“We have worked hard to develop a machine learning technology platform that delivers tangible benefits to the broker and end client,” he said in prepared remarks. “With extensive development and testing completed, we are confident in the success of implementation with Atrium.”


Lloyd’s Lab’s main effort is an InsurTech accelerator program. Atrium and Tautona first planted the seeds of what became their partnership during the program’s 2019 session.

“During the lab, we decided on very small and discreet proof of concepts so that we could look at one particular idea,” Hopper said.

The plan was to take time on the concept and develop it carefully, working on an initial proof of concept project.

After some positive progress, Hopper brought that early data to the Atrium claims team and they decided to move forward on a larger test, with other claims team members offering insight and expertise. Atrium’s IT and compliance teams also got involved, helping to integrate the Tautona technology into some of its internal systems, developing APIs, focusing on making operations run smoothly and sharing ideas.

Going bigger

The larger project focused more on blending Tautona’s cognitive automation technology – natural language processing and natural language understanding – with the broader claims adjustment process.

“They had some really good, strong experience among their team, and they were able to bring that,” Hopper said. Atrium, in turn, brought to the table “an understanding of how the property and casualty market works, how the policies are constructed, what forms are used, and how the claims adjustment process works.”

The goal now, according to Hopper, was to “try to marry the two together.”

“Teaching” the system, which communicates with uses via email and a chat function, began in earnest.

“What we did was we used some historical claims information to train the system, so we could go through and give examples of how certain claims had been handled in past policies – analysis and policy outcomes,” Hopper said. “It was a very interesting process, and we’ve been very pleased with the accuracy [north of 90%] that it’s been able to return at the end of this process.”

The process is an elegant one.

“We have a quote/bind system which has policy information in it, and it’s used from an underwriting perspective. We then also have claims modules that hang off the back of that,” Hopper explained. “The Tautona engine is able to pull information from this quote/bind system. It can pull out policy information, including straightforward things like the name of the insured and the policy period, policy referenced and things like that, but also the forms that make up that policy … it’s looking at those policy forms to be able to understand the content.”

That learning improves over time, he explained.

The partnership and its implementation took about three years to complete from idea to testing and launch. According to Hopper, the timeframe was just fine because of the unique nature of the companies’ collaboration.

“Policies themselves really can, in some circumstances, be quite complicated … they have a significant number of pages and a significant number of forms. There are different challenges that need to be addressed before you can do what we’ve now achieved together on this project,” Hopper said. “What has happened is we’ve continued to work together to improve the system as it goes.”

“The biggest benefit is working with a partner that was able to bring some really good technology and expertise in those areas that we didn’t have,” he added.

Related Stories

Keep up with the latest news and events

Join our mailing list, it’s free!