5 Questions for insurance companies looking into Gen AI

McKinsey just published their 2025 Global Insurance Report. It is an excellent reading, and you should check it out.

The bottom line is…

• … P&C is growing because people are paying more for their policies
• Home insurance costs are increasing, and sky-high house prices, absurd repair costs, and reinsurance expenses are to blame.
• Car insurance gets the volume discount treatment (aka economies of scale).
• Electric and self-driving cars are shaking up insurance faster than a Tesla in “Ludicrous Mode.”
• Insurance companies invest in risk management and mitigation to deal with Mother Nature.
• The population is getting old, so insurers pivot into retirement and health-and-wealth, and
• Gen AI is ready for prime time.

GEN AI’s Prime Time

No surprises there. Gen AI has been one of the top 3 topics in insurance over the past couple of years.

It is still fascinating to see how fast it evolved from promise to working and delivering results. Here are some examples mentioned in the report:

• Algorithmic Underwriting: Insurers are automating underwriting processes using Gen AI to enhance accuracy and efficiency in risk selection, significantly reducing time from days to seconds in some cases, such as in terrorism risk assessment, where data cleansing and geocoding are automated.

• Enhanced Chatbots: Gen AI implementation includes chatbots that provide real-time responses in underwriting workflows, improving the efficiency of handling customer inquiries and policy questions.

• Personalization and Segmentation: Insurers utilize Gen AI to develop highly personalized offerings, effectively implementing “segmentation of one,” where policies are tailored to individual customer details based on insights derived from historical data and algorithms.

• Claims Processing: Gen AI has improved claims accuracy and workflow prioritization by assisting adjusters with decision-making, thereby reducing the time spent on document reviews and increasing overall operational efficiency.

• Data-Driven Risk Assessment: Insurers are leveraging Gen AI to analyze diverse data sources (e.g., climate data, satellite imagery) for more precise risk modelling and underwriting decisions.

My own 2 cents

One thing insurance companies learned from pursuing their strategy in data analytics was that their technology and business models require a concerted effort to deliver any meaningful change and that this change takes a long time to materialize.

By the same token, capturing the Gen AI opportunity requires, before anything, starting with the right mindset.

Here are 5 questions you should ask yourself before embarking on this journey:

1.“How strategically am I thinking about this topic?”

Given the broad scope of Gen AI, insurance companies need clear priorities to focus their efforts.

This is true for all companies but even truer for small and medium-sized players, which do not have the deep pockets of larger companies and whose technology is typically less prepared.

Implementing a Gen AI-powered tool just because the NPV is positive or because “other insurers are using it” produces lacklustre results and missed opportunities.

On the other hand, identifying which AI capabilities are critical to delivering on your strategy or, even better, which new strategic goals you can achieve by committing to a specific new set of achievable capabilities will allow you to differentiate and thrive.

For example, AXA uses Gen AI to create more competitive insurance products and redefine their target markets.

2. “How 4-dimensional is it?”

Chances are the full solution to the strategic problem you solve does not exist yet.

This is the reason why so many companies shy away from bold, aspirational goals in favor of more down-to-earth but bland objectives.

However, the magic happens when you develop a portfolio of initiatives capable of delivering partial solutions in the short term while publically committing to a valuable, unsolved problem in the long run.

For example, there are several initiatives (both capitalizing and not on Gen AI) that insurance companies can adopt that alleviate the problem that buyers find insurance intimidating, but a final solution does not exist.

However, committing to solving this problem not only gives the company strategic clarity but also allows it to “own the problem.” Owning a problem comes with a number of benefits, such as customer self-selection, changes in expectations and interactions with customers, talent attraction, and organizational culture, among other things.

One example is Root Insurance, which focuses on solving the problem that insurance rates do not perfectly reflect driving behaviour. Their Gen AI investment portfolio prioritizes solutions that provide significant gains towards this problem, from driving pattern simulators to customer education.

3. “How innovative is my frame of reference?”

When insurance went online, insurance companies tried to replicate their offline processes on their websites. That did not work, and companies had to go back to first principles and redesign their processes for the new channel.

Likewise, insurance companies who try to apply Gen AI to how they conduct their business today also find limited success.

Once again, insurance companies must go back to first principles and redesign your processes end-to-end based on Gen AI’s capabilities and your strategic priorities.

One example of a company doing this is Lemonade. The company uses Gen AI to simplify claims processing, making it instantaneous by processing claims while generating real-time responses for customers.

4. How prepared is my risk management?

New world, new risks.

For each example of success using Gen AI, there are lists of failures. From oversimplified or misinterpreted legal jargon leading to customer confusion and legal disputes to excess false positives in fraud detection to the writing of custom policy clauses that were unclear, unenforceable, or legally non-compliant.

To address this issue, Risk Management and the company’s risk culture need an overhaul. Starting with education so you can make informed decisions considering both “known unknowns” and “unknown unknowns”.

A clear governance framework for Gen AI usage, with precise explainability requirements for Gen AI model and mechanisms for ethics & reputation impact reviews, is critical.

Finally, and most importantly, the you must update your risk strategy statement to reflect Gen AI risk appetite as well as how you think about risk-reward trade-offs.

5. How do I think about success?

This may sound simple, but it is far from.

You may define success as achieving specific project KPIs such as efficiency targets, error metrics, or customer satisfaction. The advantage of using these is that they can be designed to be specific and relevant for the business. However, when dealing with emerging technologies, these are often short-sighted and restricting.

You may set learning goals and milestones, such as discovering the best way to interact with your customers. These can be relevant for the business and provide both specificity to deliver impact and flexibility to foster experimentation. Learning is also the best way of de-risking innovation. However, too much emphasis on these goals (as opposed to delivering) can create risk aversion and deteriorate accountability.

Platform improvement goals such as cultural transformation, readiness to respond or adopt Gen AI, and ways of working are helpful, but they can be too distant from the business for leaders to maintain focus over long periods of time.

The best definition of success is, of course, the one that works for you and is aligned with your strategic goals. The limitations of your choices should be viewed as issues to be monitored and managed rather than problems to be avoided.

I would love to hear about your initiatives with Gen AI and your thoughts on how to make it work.

Please let me know on the comments or contact me.

Best,