The insurance industry stands at the precipice of transformation where AI could enable a fundamental shift from reactive coverage to proactive risk prevention. This evolution from traditional insurance to “assurance” represents a future where risks are actively mitigated rather than simply insured against.

What regulatory pitfalls must insurers navigate, given that the industry operates by gathering and analyzing sensitive personal customer data? Above all, how do they execute this while innovating with new technologies?

Data, analytics, and artificial intelligence can help immensely with operational efficiency and better customer experience in the insurance industry. Provider of AI-powered insights Concirrus’ CEO Andrew Yeoman even believes the notion of assurance is possible when AI-assisted underwriters make informed decisions to mitigate possible future risk.

Then, there is the computer vision startup, Cape Analytics, that is grounded in present and live geospatial data and transforms it into actionable insights for insurers to write better policies and inform homeowners what preventative measures they can take against wildfire damage, according to NVIDIA’s Global Director for Financial Services, Kevin Lewitt.

This is the essence of assurance.

Tune Protect’s approach to AI and data regulations

Insurance is a knowledge-based industry that involves processing vast amounts of structured and unstructured data. This was evident from a conversation with Tune Protect’s Group CTO, Prasanta Roy.

Prasanta Roy, Tune Protect Group CTO

Because it does not yet have a license to offer insurance in certain countries, the organization utilizes a reinsurance model which requires careful data sharing and risk allocation agreements with local insurance partners.

Tune Protect has to comply with varying data residency regulations across multiple countries they operate in, and Prasanta noted that some countries have stricter insurance regulations than others.

Tune often centralizes claims processing, either doing it itself or through a claims settlement partner, to gain efficiency depending on where the markets are operating.

“AI uses a lot of synthetic data when it comes to claims or other areas of insurance. So, AI still has to evolve much more and it’s still some time away from when it can replace a human adjuster.” – Prasanta Roy

However, the customer data still must be handled in compliance with local regulations whenever a sale has been made or a claim has been filed, with some countries requiring the data to remain within their borders. To ensure compliance, Tune works closely with local partners and regulators to determine the appropriate data governance model to apply in each market, he said.

AI in insurance – human in the loop

RPA (Robotic Process Automation) is widely used in the insurance sector to automate various processes such as claims processing, finance, and customer service. Prasanta said, “And now RPA has gotten much more intelligent with the power of AI in the background.” Tune Protect adds machine learning capability into the claims process.

Prasanta also recognized the benefit of using AI for cybersecurity as it efficiently helps to identify security vulnerabilities and false positive alerts through correlation and analysis.

For customer support, he observed the immense power of AI-based sentiment analysis on customer calls to understand the customer’s emotional state and provide better support.

But it is the AI-powered computer vision applications that really capture the insurance industry’s imagination.

With vision AI, Texas-based company Control Expert identifies vehicle information such as its make, model, color, and license plate. According to NVIDIA’s Kevin Lewitt, “ControlExpert also developed AI models to segment visible vehicle parts and precisely detect the severity of the damage, generating a detailed description of the damage as well as cost estimation for repairs.”

When CXposè spoke with Prasanta, he shared his observations of AI used in claims management. However, he challenged that AI is yet to mature to the stage of detecting the age of dents, for example.

“AI uses a lot of synthetic data when it comes to claims or other areas of insurance. So, AI still has to evolve much more and it’s still some time away from when it can replace a human adjuster,” he said.

Breaking free from Pilot Purgatory

Countries like Singapore already recognize the potential AI has, and its regulator, the Monetary Authority of Singapore (MAS), actively developed a risk framework in partnership with insurers to ensure responsible use of generative AI (Gen AI).

Regulatory sandboxes can also simulate live market and regulatory environments for participants to assess how their products and services would perform. Last year, Tune Protect participated in one to test their appetite in offering affordable life insurance to the underserved B20 income group. It used AI and analytics to develop a model that determined the affordable coverage plan.

A more holistic reimagining of domains like claims, underwriting, and distribution can actually better serve the AI initiative in an organization.

With so many promising and demonstrated applications for AI and analytics in the industry, it seems impossible for pilot purgatory to happen. And yet, this is what McKinsey pointed to in a podcast between three partners, when one of them said, “…a lot of time is being spent on testing, analyzing, and benchmarking different tools such as LLMs.”

Organizations also tend to prolong the pilot because it does not generate the kind of value they hoped. But many pilots have little impact due to the isolated nature of the use cases, so a more holistic reimagining of domains like claims, underwriting, and distribution can actually better serve the AI initiative in an organization.

Meaningful change may mean redesigning workflow and data flow processes. “This approach requires investments in more than just tech,” another McKinsey partner, Khaled Rifai, said.

Another factor that could be holding back pilots or proof-of-concepts from going live is data.

Winnie Chua, PolicyStreet, co-founder

In an email interview with Winnie Chua, co-founder and Group Chief Product Officer of PolicyStreet, she shared that while early AI steps have been made by insurers in Malaysia, “…Malaysia remains some distance from achieving a level of data consolidation and processing that could meaningfully transform how insurance coverage and financial services are tailored for users.”

Recognizing the transformative potential that AI holds for insurance domains, Winnie did add that a key focus for PolicyStreet this year would be advancing its capabilities in AI. “As the insurance and insurtech sectors work to ethically consolidate sufficient data and develop AI capabilities to process it, we anticipate significant advancements in underwriting processes within the market.”

The future of Assurance

The real promise of AI in insurance isn’t just in making existing processes more efficient – it’s in fundamentally transforming how we think about and manage risk.

By leveraging AI’s potential to pre-empt while respecting data privacy and regulatory requirements, a future where risks are prevented rather than just insured against seems the natural goal for insurance to evolve into.