AI: insurers must mimic Google, Facebook and Amazon
By and large, insurance companies are digital immigrants, ie, they traditionally did not leverage digital and data to create competitive capabilities. To transform they must learn from digital natives such as Amazon, Facebook, Google, etc.
These companies have been able to democratise advance analytics and digital across their value chain. Artificial intelligence (AI) and machine learning are a natural extension of advance analytics and digital. Many insurance companies have a philosophy to have a centralised group for AI. This is a short-sighted approach—although it is fair to start with a centralised team to start the capability, the vision must be to democratise the skills and appreciation for advance analytics, machine learning and digital.
Unless a company creates a strong training programme, a lab environment for controlled experimentation, and embeds the adoption to performance ratings at all levels, the true innovative potential of intelligent automation will get diluted under corporate bureaucracy and politics.
That is the view of Abhishek Breja, now a consultant, and the former head of performance transformation and AI at Assurant.
The case for intelligent automation
Breja joined Assurant in 2009 as a global delivery model and performance transformation expert, to the company’s management committee. His role was to help the leadership in building a roadmap for transforming the company from a decentralised organisation to a centre-led organisation, that could create significant shareholder value by making operations more efficient through consolidation, and standardisation of its operating model for operations.
After leveraging and fully exploiting all other traditional tools for transformation and operational efficiency, in 2016 Breja championed the case for intelligent automation to get incremental and non-linear transformation benefits.
“By 2015, we had already exploited the traditional tools of outsourcing, consolidation and standardisation to a great extent. In my opinion, trying to get more savings using the same tools would have been penny wise and pound foolish. Therefore, I had to think creatively about other ways to deliver more efficiency,” he says.
At the time, robotic process automation (RPA) was a new technology everyone was talking about. After a few initial, not so successful, experiments with RPA Breja started acquiring technical knowledge on machine learning. Over time he reached the conclusion that the most effective way of creating true tangible value is to develop initiatives that leverage the combined strengths of RPA, machine learning and reimagining end-to-end processes.
“We started to reimagine the potential of the process,” Breja says. “When you combine these things, the results are transformational. It can be used not just to reduce cost but to create a whole new experience for customers, as well as influencing your product strategy.”
Central to the success in the company, he explains, was the idea that this technology should not be limited to a unit or a silo but embraced by the entire company.
“Amazon and Facebook and Google do not have a centre of excellence, they embrace it throughout their companies,” he says. “I wanted to democratise the potential of intelligent automation by enhancing it with machine learning and make it available wherever it could be of value.”
He admits that this is not always easy within insurance companies. A gap can exist in the language and understanding of data scientists and technology experts compared with underwriters or claims managers and other executives within insurers.
“You need translators, techno-functional people, between the two to make things really happen,” he says.
Breja will be speaking at the Intelligent Automation in Insurance conference in London this April. Find out more here.
Application of the technology
Assurant applied the logic to what Breja calls the low-hanging fruit and ended up applying it to four main areas.
The first was any sizable and highly rules-based process done manually. “You might have individuals carrying out a 32-step process that could be automated with the right system,” he says.
“That is the most obvious application for insurers. However, what may seem as rule-based to human may not necessarily be ‘codifiable’ for a bot, it may require some basic common sense that we humans take for granted. That’s where basic machine learning comes into the picture.
“Additionally, the process was designed for humans, and there may be obvious changes that would work better for machines.”
The second application was in areas where a company might not currently operate because it was not economically feasible to do so with staff members doing the work. He gives the example of some areas of fraud analytics where the potential savings versus the cost and the level of errors mean that it has not been worth the money.
“AI can make some of these processes economically viable,” he says. “Plus, the accuracy is much higher. The results can also be fed back into the underwriting process to reduce errors and improve margins on that side of the business as well.”
The third application, Breja says, is large scale IT modernisation projects that are underway. Many insurers have legacy systems, many of which are being modernised. Chief technology officers (CTOs) are under great budgetary and timeline pressure on these projects because most of them are behind schedule and significantly over budget. That also means innovation suffers.
“We worked out that there is a percentage of ‘functional points’ these projects aim to automate through a traditional IT system development approach, that can be speeded up by the use of intelligent automation.
“Selectively there will be functional points where intelligent automation may get you 80 percent of the ideal desired functionality at 50 percent of the cost and five times faster. This can be a huge help in relieving the pressure on IT budgets and project timelines,” he notes.
Finally, he says, it can be placed at the centre of growth and product development in insurers. It can be used to help develop policies and, more importantly, add a cutting-edge customer-facing layer to new launches that can allow bigger companies to compete with fast-moving startups focused on delivering a differentiated customer experience similar to that offered by companies such as Lemonade.
“Many insurers are increasingly worried by companies such as Lemonade that are grabbing market share quickly because of the way they engage with customers,” Breja says.
“By using intelligent automation insurers can improve their products quickly too and create a customer-facing layer that can set them apart despite the rigidity of their back-end system which is painfully slow to transform and modernise.”
He admits that big variations exist in the speed at which insurers are willing to use things such as intelligent automation, and the way they do it, and he believes that companies that adopt it more aggressively will gain a significant advantage over their rivals.
Breja’s advice
Start small but be bold—in the hope that RPA is a silver bullet, do not shy away from using basic machine learning and process re-imagination.
Remain value-centric from day one—do not automate for the sake of demonstrating the potential of automation and disregarding the business case. Process fragmentation and the future complexity of maintaining a digital workforce may easily erode all your dreams of the pot of gold at the end of the rainbow.
Democratise intelligent automation—the potential for intelligent automation cannot be fully leveraged by a CEO, it needs enterprise-wide adoption. Empower your company employees at all level to play with intelligent automation provided they take accountability for a strong return on investment for each initiative.
Abhishek Breja will be joined by other industry experts in an unmissable panel disucssion at Intelligent Automation in Insurance on April 26 in London, detailed below. Getting the Basics Right: Design Thinking & Re-Imagining Your Process Achieving the basic needed environments with AI always on your mind
