'A' for AI, 'I' for insurance (2/3) - how AXA ranks human intelligence alongside its artificial counterpart

As it enters its fifth decade, AXA re-affirms its original 1980s mission statement to re-invent insurance. In 2026, that takes the form of a strategic plan called Unlock the Future and, of course, AI plays a large part in this.

CEO Thomas Buberl cites the AI thinking as another example fo what he pitches as AXA’s “entrepreneurial DNA”. AI will, he argues, transform the insurance business for the benefit of customers, as well as drive internal organizational change:

Our employees today are highly involved in the use related to AI in our business. A recent internal barometer shows that 70% of our people are optimistic when it comes to the future benefits coming from AI. Thanks to AI, we are moving forward the major levers of insurance for better risk pricing, better service quality and better customer experience, speeding up of technological developments for our coders to bring on board our people.

And Buberl is commendably focused on the human intelligence aspect of all this:

It's not a technological challenge. We are investing very strongly in our employees' training. It is our duty because we do know that our success depends on their -- them embracing our strategy and using these new tools and the trust in these new tools.

Re-invention

AXA also sees a “very strong opportunity to pioneer the way in the insurance industry on how to utilize Artificial Intelligence solutions, boasts Guillaume Borie, the firm’s Global Head of Finance, Strategy, Underwriting, Risk & Technology:

In the last few weeks, we've heard that for incumbent operators like the insurance companies, AI is seen as a challenge. But for us, we see AI more as an opportunity, as a unique opportunity to re-think the way we inter-operate with our customers.

AXA has the assets to pull of a major transformation here, he argues:

The first asset is the robustness and quality of our distribution channels, which will remain a key pillar for us and which now on the back of AI will have the opportunity to be more efficient, more productive and to focus more on what they do best, i.e. being on the front line with their customers. And we have a great example of that in France with the deployment of a chatbot, which is being used daily by all of our agents in France. Our agents and their teams can better support and service their clients when they come to them for advice on an insurance-related product.

The second competitive advantage we have to successfully pull off this change is our scaling up momentum. We have pockets deep enough to massively invest to use agentic AI to scale it up across the group and to generate concrete efficiencies.

Currently only four percent of AXA’s tech budget goes on capital investments, the overwhelming majority being allocated to ‘keeping the lights on’. That has to change:

We are determined to increase our investment budget because the investment needs into AI will obviously be increasing...What’s important to note is that this growing investment is a real opportunity to start laying down the foundations and future vectors of growth going forward.

AXA’s “very attractive brand” will help it achieve its AI transformation ambitions, he suggests, pivoting back to the importance of human intelligence as a driver:

In this transformational change, our capacity to attract the best talent will be a critical success factor. In the last few months, we've recruited more than 900 data scientists as well as data engineers. And today, they are working across the world to improve our data models and to develop AI solutions, services and applications, which are gradually being used by all of our experts.

To succeed in AI, the first thing you need is fundamental technical and technological knowledge and skills. Machines don't learn by themselves. We have thousands of experts who are specialists in their field who know all the ins and outs. And the rules we apply day in, day out are a key asset to move faster and to pioneer the way in this transformational change.

Use cases 

To date, there are two main fields of AI implementation. The first is client-facing, explains Buberl:

In other words, how can we really improve the relationship and the interface with our clients? And here, distribution centers are key. Our idea here is to work, of course, on the automation front, on the customization of client interface in regard to service, but also help our general agents, our distribution networks, to use properly AI in order to improve customer relationship.

The second main application of the tech involves how AXA assesses insured risk and how it  can prevent it, he goes on:

An insurer has available lots of data. We are working at length about the quality of data because if you load up a car with not the right gas, it doesn't work well. That's why having good data banks and also having the ability with AI to use non-structured data [matters]. For example, if you insure this room here, the risk engineer comes around and tries to understand the danger of potential fire, he will make a PDF report that we can use once. But if you want to use it several times for other risks, you have to be able to make use of such piece of information. AI for the first time is helping us do that.

In other words, [it’s about] assessing a risk on the one hand and then the question of prevention in other words, what can we do to avert the next claim will be much more specific and precise. And these are the two areas on which we are working as it comes to the implementation of AI. 

In terms of business use cases, Patrick Cohen, CEO of AXA’s European Markets & Health operation, picks up the story:

To be very concrete, with respect to growth, and expansion today. AI makes it possible for us to have an underwriter prepare their work so that they can refine their offers and provide more offers, more proposals in a shorter lead time, getting information from the claims from the various brokers, from the agents, processing more funds to generate more growth.

Another interesting example is that the AI solution will give us the right offer with the right price at the right time to the right clients. We are using agentic AI, which in a predictive manner are able to identify, detect needs and basically automatically create e-mail and correspondence and communication, which are totally customized to fit the individual situations of the clients. Very interesting.

The last example is how we'll be training and developing our agents, providing them with advisory tools in their client interactions, and we'll be ramping up their skill sets and expertise on our products. We'll be very quickly generating videos, tutorials and training sessions for use by our agents, so much for growth.

There are also opportunities to enhance client servicing, he adds:

AI is now capable of solving a great number of pain points which you have in the field of insurance. Obviously, one of them is not to repetitively have to describe your events when you have incurred a claim and you report it to also radically shorten the lead time for reporting in a seamless process where you do not have to repetitively send a piece of document. This is being done in Germany, Italy and Switzerland at scale with a very tangible example.

Today, in our claims management centers, the claims handlers now have what we call an assistant, which basically put them into cruising mode, ie, when they interact with clients, the AI tool will suggest an answer, will get the right data at the right time so that the claims managers focus on empathy on better serve the client and speak with them, listen to them, satisfying the clients better and making them more effective, more productive.

Impact

What about AI’s impact on the wider competitive landscape? AXA has a 40 year history, but that doesn’t guarantee a continued place in the leadership of the sector. CEO Buberl has a Gallic nonchalance here when he argues:

Of course, AI will create newcomers. In the digital revolution that we saw several years back we saw newcomers come along. Some of them survived, but most of them either did not survive or chose an interesting fate, which is instead of competing with insurance, they created a partnership with insurers so that they can work with them together.  I think the same thing in most cases, should occur again

He concludes with what is essentially an invitation to newcomers and start-ups in the field:

The insurance model is a scale-based model, but we need to modernize it. We need to simplify it, automate it and customize it, the customer interface and innovative companies based on AI and on the LLM models will be helpful. This is why, in order to do better this time compared to the digital revolution, we must, from the very outset, try to find partnerships to benefit together from the positive effect of AI on our processes, on the customer interaction. And along those lines, then, of course, we are working with all the model providers, in particular, with Mistral.

My take

Next up, Admiral.