<img height="1" width="1" style="display:none;" alt="" src="https://px.ads.linkedin.com/collect/?pid=4958233&amp;fmt=gif">
5 min read
Robert Anderson

When ChatGPT was first released, it took the public by storm. Almost simultaneously, AI image generators were also released to equal acclaim. Suddenly, it seemed like generative AI (Gen AI) was capable of doing anything and everything. People everywhere were wondering how they could use it to make their lives easier.


And then, with the increased usage, cracks started appearing in the generative AI solutions. ChatGPT and other large language models (LLMs) were pulled up for getting things wrong or making things up (hallucinating) if they could not find the information they needed. Image generators also couldn’t do certain things correctly, such as recreating hands or text. For illustration, here’s an image I created just now of a bicycle race across the Sahara Desert – take a look and see if you can notice what’s wrong…


Bicycle_Race_across_the_SaharaImage generated by DALL-E 3


As often happens, disillusionment and cynicism started setting in. It became easy to recognise the signs of generative AI content. For instance, images had a glossy sheen that made them feel slightly surreal. Or, text was too formal and factual and without personality or style.


But increasingly, we learnt, as the AI did, that accurate prompting was essential and that AI content could be shaped by asking it to adopt an approach or specified persona.


Reminding us that generative AI is on a journey and learning at a velocity we probably can’t comprehend.


Not just insurance automation but experience


So, is there a meaningful way to leverage AI in the insurance industry? After all, it’s already being leveraged in many other industries. And does it have a place in customer service, the most critical interaction space we have with our customers? The answer is a little more complex and nuanced than a simple ‘yes’ or ‘no’. There is definitely a place for it, but, as we have learnt with AI prompting, you have to do it correctly or your customer experience could become a differentiator, but not in a good way.


Policy administration


Customer service during policy administration is a key moment of truth, especially when it comes to inception, mid-term adjustments (MTAs) and renewals. It influences perception and loyalty, which is difficult to build but very easy to lose – whether that dissatisfaction comes from long waiting times, multiple transfers or inconsistent answers.


One way in which generative AI can help enhance the customer experience during policy administration is by quickly providing customer service agents with the correct and relevant information so they can provide that personal touch. This can cover things like policy inclusions and exclusions, policy limits or more personal data. For instance, we know of at least one insurance company that is using generative AI to provide the current stats of a customer’s favourite sports team to help build rapport and trust.


It’s best practice to exercise caution when using AI to provide direct services to customers. We have likely all experienced the early use of chatbots. Although they have come a long way, in their early days they were inflexible and often lost context, causing frustration when the user had to start their query all over again. Since then, chatbots have been improved through better training and better support for natural language processing (NLP). Now some insurance companies are using chatbots to provide more detailed answers where the information is easy to glean from the back-end systems.




Underwriting is often a complex and time-consuming process that involves multiple data sources, rules and models, especially in speciality insurance.


Generative AI can help streamline and improve underwriting by extracting and validating relevant information from structured and unstructured data sources, such as submissions, annual reports and third-party databases.


Generative AI can summarise and categorise the huge amount of data to help underwriters to generate risk scores, pricing and terms. The term co-pilot has come about, which depicts AI as an assistant, there to help and augment the underwriter’s role.


In speaking with risk managers at insurance companies, we’ve learnt that they don’t like being unable to explain the decisions the companies make. But you can, in fact, counter this with generative AI by using it to help explain the underwriting decisions and provide suggestions for improvement.


This technology can therefore help to increase the speed and accuracy of underwriting, reduce operational costs and offer more personalised and competitive products to customers.




The claims process is one of the most critical parts of the customer journey, which usually happens during or just after a stressful event. In fact, the claims process is arguably the product you’re selling, whereas the policy simply defines the rules of engagement.


However, many claims processes are still manual and inefficient. Generative AI can help automate parts of the claims process by helping to provide insights, identifying patterns and outliers and detecting fraud. Endava has already built an insurance fraud detection system for a customer that was based on pattern analysis and relationships, looking at a person’s known associates for any risk.


Recently, a head of claims told us about the issues they were having providing their field adjustor teams with the right information. Once trained, generative AI can analyse this field claim data, verify the policy coverage, assess the damage and even estimate the repair cost. AI can also provide recommendations to adjusters to help make their estimates quicker and more consistent.


In turn, insurance companies can reduce the claim cycle time, improve the accuracy and consistency of claims decisions and ultimately enhance the customer experience.


Nurturing potential


Generative AI has exciting potential and, I’m sure, will become an essential tool that insurers will need to embrace to gain its many benefits. It also learns with deliberate consistency. Problems we’ve already encountered are overcome as AI iterates with finer and finer strokes.


By efficiently managing tasks and freeing humans from the mundane, we can foresee generative AI helping to improve the customer experience. Allowing us to make the all-important human connection where it is most needed, whether that’s providing much needed personal support during a claim or responding to account queries with the rich data that AI can so efficiently serve up.


To learn more, download our e-book, Leveraging data in insurance: developing an effective data and AI strategy, or get to know our first-of-its-kind AI solution that combines the power of data and multi-agent autonomous teams.


No video selected

Select a video type in the sidebar.