Four ways in which AI is transforming insurance
Artificial intelligence has the potential to reshape the sector by streamlining and even totally automating key processes, while providing insights that can help insurers to make more accurate risk assessments
Insurance companies often find it difficult to price risk accurately. The risk profiles that they create are limited by the data they have on customers, which tends to be generic (age segmentation, for instance). This means that the premium a policy-holder pays won’t always match the risk presented by that individual.
By using AI, insurers can potentially offer more accurate pricing because the tech is able to process more data points, notes Nelson Castellanos, chief partnerships officer at HDI Embedded.
“Why should you pay the same price for a policy as someone else if your profile is less risky than theirs? AI allows for more dynamic pricing based on the actual risk,” he says.
The latest AI systems can do this by processing vast quantities of unstructured data, as Ed Halsey, co-founder and COO of Taveo, explains.
“There’s a lot of untapped data sitting in filing cabinets on old paper documents or pictures stored on someone’s hard drive, but you don’t know what they’re showing unless a human eye looks at it,” he says. “If you can get all that information into an AI tool, it will make underwriting so much more efficient.”
One way for insurers to guarantee a poor customer experience is to let the claims process drag on. AI can help to solve this efficiency problem by automating much of the work required. HDI Embedded, for instance, uses AI to read documents such as police reports in seconds, removing the need for a human to read them – a potentially onerous and time-consuming task.
Another way in which AI can accelerate the claims process is by automating damage assessments, says Simon Thompson, head of data science at GFT Technologies, a digital transformation consultancy.
He explains: “If you took some photos of your vehicle that showed no damage when you took out your policy and you then send in pictures showing big dents and scratches, it’s easy to have a machine that says: ‘Yes, significant damage is present in those images and therefore there’s no debate that there was a claims event,’”
This means that some claims can be processed without any human intervention. “The machine learning model will sometimes come back and say that there’s no significant damage,” Thompson says. “In such instances, you would need a human to intervene and manage any disputes arising, but the process can be completely automated in most cases.”
Advanced generative AI tools such as ChatGPT can provide valuable insights to employees and customers alike, according to Fady Khayatt, partner at management consultancy Oliver Wyman.
“There’s an opportunity to use generative AI as an aid to decision-making by treating it as a co-pilot. It’s not removing the human from the process; it’s providing the human with additional information to support better decisions,” he stresses.
For instance, AI can provide recommendations based on similar cases, with the claims handler or underwriter then validating that information.
Insurance firms such as Esure are also using chatbots to make it easier for customers to obtain answers to basic enquiries, potentially resolving these more quickly and efficiently than other methods would allow.
If a question can’t be answered satisfactorily by a chatbot, it will send a summary of the text conversation to a member of staff, says Alison Edge, data product owner at Esure. This means that the customer won’t have to repeat what they’ve typed and the employee can comprehend the query faster than they otherwise would.
“A chat history can often be extremely long, so summarising it means that customers won’t have to wait for the agent to read the whole thing,” she says.
As many as 40% of commercial properties in the UK are underinsured, according to research published by global insurance broker Gallagher in Q4 2022. Thousands of policy-holders could therefore find themselves seriously out of pocket if they need to make a claim.
“Many properties are insured at what the rebuild cost was 30 years ago and we’ve just applied index-linking every year,” Halsey reports. “That’s not adequate – things change.”
AI can help to solve this problem by improving the accuracy of building insurance coverage, he says. It does so by combining geospatial data and satellite imagery with AI-powered computer vision, giving a more up-to-date view of a property’s dimensions and risks.
Insurers can go further by using drones to assess the risk profiles of commercial buildings, Halsey adds.
“You can send a drone around the top of a property to assess the condition of its roof. Or, better yet, you can get an agreement for it to fly through a warehouse, say, see all the risk features and tag these with computer vision,” he says. “All of this sounds like it’s the year 3000, but it isn’t. You can do it now.”
What are the insurance industry's digital transformation priorities?
How much are customers driving digital transformation and what are the investments priorities for insurers?
How insurers can harness the power of generative AI
By adopting GenAI, insurance firms can become smarter and more productive while pushing standards of customer service to new heights. But achieving all this will require a change of mindset
The rapid advance of generative AI has the potential to reshape the insurance industry from top to bottom. Largely manual activities, ranging from risk pricing to claims handling, could be enhanced or even completely automated by firms adopting the latest GenAI tools.
“These are the lowest-hanging fruit when it comes to applying generative AI – and they’re also the lowest-risk uses,” says Rahul Kumar, vice-president and general manager of financial services and insurance at Talkdesk. “You’re not putting generative AI directly in front of your customers and you have the opportunity to test, learn and optimise by using the technology internally first.”
For instance, GenAI can help customer service agents become more effective and efficient by accelerating document searches and image scans. It can give them real-time guidance while they’re interacting with customers, Kumar adds. It can even help new agents by generating scripts to follow based on the conversations they’re having.
While GenAI can enhance internal processes, it can also be used to take the customer experience in new directions. For instance, the technology could dynamically adjust a customer’s coverage as their circumstances change.
“Imagine a scenario where you’re going on a long car journey,” Kumar suggests. “You enter the destination on your satnav and your insurance co-pilot suggests increasing your cover, because it can see that you’ll be travelling through territories more prone to things going wrong. That’s a new type of customer experience because it’s the insurer proactively looking out for you.”
Despite the potential that GenAI offers insurers, integrating it into their systems and processes won’t be straightforward, partly because of legacy technology problems, but also because players in this sector generally aren’t quick to embrace change.
“If you compare insurance with other sectors, its technology adoption has traditionally been a challenge because the industry is very risk-averse,” Kumar says. “Insurers typically take the fast-follower approach: we need 15 people to tell us that something new is working for them before we even consider talking about it.”
To embark on the process, insurers must take numerous steps. First, a mindset shift is required: insurance leaders need to become more open to adopting new tech. Second, they have to recognise that integrating AI is an iterative process, so they shouldn’t be afraid of getting things wrong as they test and learn.
“If you don’t incentivise people to fail, you’ll never get them to try out new things,” Kumar stresses. “This is about promoting an organisational culture where it’s OK to say: ‘This might not be perfect on day one, but we’ll get there eventually.’”
Insurance firms must also ensure that humans are heavily involved in any AI processes, particularly given the risks concerning so-called hallucinations – where GenAI makes fictional claims that it confidently passes off as facts.
“You need to put guardrails in place to ensure that the technology doesn’t go rogue,” Kumar warns. “Automation is great, but you must have a human in the loop so that they can see what’s going on and intervene if things aren’t proceeding the way they should be.”
Humans also need to ensure that GenAI tools are being trained on accurate information, he adds, pointing out that “the insights that generative AI outputs will only be as good as the data it has access to”.
As regulated entities, insurers have to be cautious about potential compliance problems concerning data protection, particularly when they’re using third-party tech providers.
“Proper control mechanisms need to be in place when it comes to who can access and analyse data,” Kumar says.
A final tip for ensuring a successful GenAI deployment is not to try to implement everything in one shot.
“You don’t have to go all in with a big bang,” he says. “Conduct internal tests first with your employees, picking a few processes where the risk isn’t that severe but the possible efficiency gains are significant.”
In other words, focus first on areas where using GenAI can achieve cost savings and then move to areas where it can generate value, such as customer-facing processes. By taking that approach, insurers can radically redesign their businesses without causing unnecessary disruption.
While a hasty implementation project will undoubtedly cause problems, such is the potential offered by GenAI that insurers must still act quickly and decisively. Indeed, Kumar’s prediction is that, “if things continue to go in the direction they’re going, generative AI will transform the industry to the point that it will be almost unrecognisable by 2030”.
Lifting the service ceiling: above and beyond policies and claims handling
In an increasingly crowded and competitive market, insurance providers must go the extra mile to engage customers and provide experiences that will result in lasting relationships. Here’s how
For many people, buying insurance is a simple matter: you purchase your cover and hope that you’ll never need to use it. That means most customers’ interactions with insurers will be at the point of sale or when something has gone wrong. The latter situation is where an insurer can really differentiate its customer experience (CX) by providing an empathetic service that builds trust and loyalty among policy-holders.
“Some insurers really focus on controlling the claims process, so they take charge of the supply chain more than others and don’t use so many third-party providers when it comes to intervention and repair, for instance,” says Fady Khayatt, partner at management consultancy Oliver Wyman.
An insurer that can respond promptly to a customer’s problem – a water leak in their property, say – and resolve it quickly will potentially reduce the damage caused and also strengthen the relationship by proving its attentiveness to the policy-holder in their moment of need.
“When a customer makes a claim, it’s during a heightened emotional moment where a good interaction has a much bigger impact than it does at other points in the process,” Khayatt says. “Players that control their supply chain closely and use it to deliver good outcomes for customers can help themselves in terms of lower claim costs, but doing so also has a big positive impact on their customer experiences and policy renewal rates.”
British insurance firm Esure is taking a similar approach by using AI to process car insurance claims, automating the initial damage assessment and then recommending local repair services for policy-holders to choose from.
“That helps our customers get through the claims journey so much faster,” says Alison Edge, data product owner at Esure. “It saves cost for us, because the settlement time is reduced and the vehicle can get to the garage for repair more quickly.”
Personalised insurance offerings
Insurers can also use AI and advanced data analytics to hyper-personalise their CX so that all product recommendations made to a prospective customer are relevant and based on that person’s specific needs. For instance, when it comes to covering small firms, providers that can make their offerings more relatable to a particular customer’s business are likely to be more successful.
“When a small business owner feels that the conversation and the policy are clearly tailored to their requirements, you’ll typically see a clear uplift in conversions, as opposed to when they’re interacting with an insurer that’s offering products that sound a bit generic and could apply to any company,” Khayatt says.
Another way by which insurers can elevate their CX is to make buying cover more seamless and convenient by partnering with businesses and offering embedded insurance products when a customer makes a related purchase. For instance, they could offer a travel policy to a customer who’s just booked a holiday.
“For fintech firms and marketplaces, embedded insurance also increases engagement and the ‘stickiness’ of their platforms,” says Nelson Castellanos, head of partnerships at embedded insurance specialist HDI Embedded. “That’s because the more products and services they can offer, the more likely their customers are going to stay in their ecosystem.”
Castellanos reports that his firm sees conversion rates increase by 15% to 20% when partners offer insurance products that are embedded at the point of sale.
Streamlining customer interactions
Insurers should be considering ways to minimise the friction that customers encounter when dealing with them, according to Simon Thompson, head of data science at digital transformation consultancy GFT Technologies.
That will not only help improve the CX a firm provides; it will also give the insurer greater revenue-earning potential, because customers are more likely to choose a provider that offers the most streamlined processes, he argues.
Technological advances are also creating more potential for insurers to completely redesign their offerings by thinking differently about customers’ likely future needs and how their products can be packaged.
“For example, you could have an integrated home insurance and assistance product with a chatbot and online expertise for solving problems around the home bundled with insurance, or a tool that would help bundle the right sorts of cover for a small business,” Khayatt suggests. “So you have a real modular insurance product, with AI helping to identify the right combination of products and services to offer.”
This is a pivotal period for the industry – a time in which any insurers that are slow to embrace tech to elevate their CX are sure to get left behind. That’s the view of Ed Halsey, co-founder and COO of insurance broker Taveo.
“There’s an impending explosion of AI and automation,” he says. “The brokers and insurers that can’t work out how to integrate this technology and are still operating on legacy systems simply won’t be able to match the service levels being provided elsewhere by players that have built modern, innovative tech stacks with the customer at their centre.”