How natural processing is changing insurance

Vol: 45 Issue: 4 | December 2022

IN SHORT

•  Natural language processing is evolving quickly as it helps insurers handle claims faster and more efficiently.
• The technology frees up claims team members to concentrate on other elements of customer experience.
• NLP skills can be acquired through education and industry training events.

IAG

IAG deploys natural language processing (NLP) to help it predict if a motor vehicle involved in an accident is a total loss or if it can be repaired, cutting claims times dramatically.

‘This has enabled us to reduce the process from what was typically 15 to 20 days to about two to five days,’ says Danielle Handley, executive general manager, Customer Experience at IAG.

Acknowledging that delays in assessing damaged vehicles and knowing the outcome are a ‘customer irritant’, Handley says NLP technology, in tandem with machine learning, helps IAG make faster judgements. ‘So, within 24 to 48 hours, a customer can have a level of confidence and clarity about their situation.’

NLP and artificial intelligence (AI) are used to translate customer conversations into text and analyse and automate claims processing where possible. ‘The context of a claims situation does come through conversation, so it’s critical for us to be able to leverage that language and any insights to understand what’s going on with a claim,’ says Handley. ‘Then, we can refine and automate some of the processes.’

Significantly, this frees up claims team members to focus on more critical aspects of the customer experience. Handley says the discovery that targeted human interaction is still required during claims has been an important learning for IAG. 

‘You can’t assume that you can go from a 100 per cent human process to a 100 per cent digitised process,’ she says.

Since 2017, IAG has increasingly been using NLP and AI to help manage high volumes of customer calls, to undertake administrative tasks and to bring personalisation to digital quoting processes. The last involves using a machine-learning, real-time decision engine that can analyse voice and digital channels and predict customer preferences based on past patterns.

Other innovations are likely to include using robot-voice AI for outbound communication, as well as digital avatars on the IAG website, to interact with customers.

‘There’s a lot of opportunity with real-time voice technology for us to more accurately engage with our customers in terms of their sentiments and how to help them in the most effective way,’ says Handley.

‘That includes better analysing and predicting ways we can support them, potential complaints and interacting with customers in a more personalised way.’

Atradius

Global credit insurer Atradius co-builds customised models, using NLP and AI, that can read financial statements and annual reports and process information mined from hundreds of thousands of websites. 

Operating out of Amsterdam, with regional offices in Singapore and Sydney among more than 160 locations around the world, Atradius also uses neural networks — a form of AI that mimics the human brain through a set of algorithms — to deliver automatic credit decisions.

‘This saves us tremendous amounts of time and manpower, expands the scope and reach of the information in a fast-changing world, and brings immediacy to the way we manage portfolios and make risk decisions,’ says Stan Chang, director Group Buyer Underwriting at Atradius. 

He adds that NLP and AI technology is ‘an important driver in our business, as it frees up humans to perform complex tasks like servicing customers and selling policies’. It also delivers savings for Atradius through greater efficiency and productivity, while allowing higher-quality and faster underwriting decisions. 

Chang says better-informed, data-led decisions mean that customers benefit ‘from our increased willingness to insure their trades’. Given that Atradius processes high volumes of credit applications globally and manages a massive portfolio of credit risk exposures in hundreds of trade sectors, he says the use of technology is a must. ‘Aside from complexity, the cost of doing this well is prohibitive without technology, considering the highly competitive premium rates that we charge.’

The biggest challenge for Atradius in embracing such technology has been adoption. ‘This has entailed using unfamiliar technology to deliver a proof of concept, building prototypes that can be scaled, and not knowing all the answers to important questions from the outset to translate a new technology into useful business outcomes,’ says Chang. ‘[But] we had faith. We researched and collaborated. We were excited about what we were doing, and we received much internal support and encouragement.’

For others considering using NLP and AI, Chang advises building knowledge through education, networking and attending relevant industry conventions.

‘You don’t need technical knowledge to strategise your approach,’ he says. ‘Be purposeful and focused in your business goal rather than romancing the technology. 

When you’re ready, acquire skills if you don’t have them already — either through training, recruitment, collaboration or outsourced services, or a combination. Start small.’

The key, however, is to act. ‘Few businesses that shun technology will survive, much less prosper,’ says Chang.      

Tower

New Zealand-based Tower Insurance is fighting back against claims fraud through a partnership with FRISS, a Dutch company that provides AI-powered fraud and risk detection solutions.

Tower chief claims officer Steve Wilson says that late last year, the insurer recognised the need to pay claims faster and joined forces with FRISS to automate the process of detecting genuine and suspicious claims.

‘Many insurers have dedicated fraud teams, but it’s a time-consuming process identifying each claim as genuine or fraudulent,’ says Wilson. ‘AI complements what our humans do. By learning from our investigations, AI can pick up on systemic trends and identifiers happening across Tower claims that, as humans, we sometimes cannot see in isolation.’

The FRISS solution features AI and NLP techniques such as predictive models, network analysis and text mining. By inputting historic data on genuine and fraudulent claims, the tool can sift through new claims in seconds and detect suspicious cases. ‘Having an automated process that separates low-fraud-risk claims from the rest of the pile significantly speeds up the process and enhances accuracy,’ says Wilson.   

He adds that insurers with automated fraud-detection systems can also harness real-time data to respond swiftly as new scams are detected, mitigated and prevented. ‘It brings more transparency into how claims are processed, which simplifies insurance for everyone involved.’ 

Wilson says technology and data are at the forefront of Tower’s strategy to innovate within the insurance industry, while helping customers with simpler and more rewarding insurance experiences. He is confident that embracing the technology brings mutual benefits for customers and staff alike. ‘Customers love that their claims are dealt with fast and with less hassle, but it’s also hugely beneficial for our people and business,’ he says.  

While the alliance with FRISS is in its early days, Wilson says it is an important movefor the business. 

‘Tower is dedicated to helping Kiwis prepare for the unexpected. So, it falls on to us to ensure we have the most efficient system in place to reduce unnecessary costs and improve customer experiences.’