Perspective

AI fluency and adoption: how insurers can make their AI transformation stick

There’s a line often attributed to Einstein: insanity is doing the same thing over and over and expecting different results. As I wrote recently [link], we’re seeing past mistakes crop up again in our AI efforts. Our recent research shows that every insurer has moved an AI pilot into production, yet only 23% describe AI as transforming their business. Many firms are stuck on repeat, caught up in the stubbornness of the status quo. So how do we break the cycle?

Change is famously hard, especially when it comes to organisational culture. Leadership and cultural resistance are the number one barrier to scaling AI (39%). All those habits, behaviours, beliefs, biases, and professional and personal goals get in the way. Culture eats strategy for breakfast, as the saying goes. So what are the firms making headway doing differently?

I’d bet my house that focusing on AI fluency and adoption can break down a lot of this resistance.

Cultural resistance: people have long memories

Ask anyone who's worked in insurance and they'll be able to tell you about a time a confusing piece of tech landed: the painful onboarding, the disruption and, certainly, the cuts. If we expect people to embrace AI, or any technology for that matter, people in the industry need both reassurance and an incentive. They need to understand the why and the how. Without those, we risk making their lives harder and making them nervous about their jobs changing or, worse, disappearing.

Communicating the big why

Leadership must make clear that any AI rollout isn’t about cuts, but about growth for everyone. I spoke before about an insurer piloting AI in cyber underwriting [link]. The AI turns a submission into a quote recommendation in seconds, and an underwriter reviews every one. Letting people do the real human work, and removing the onerous and laborious stuff, must be our north star.

Leaders need to tell an aspirational story that pricks ears and sparks imagination: one that helps people see how their day-to-day job can be better. Less admin, yes, but also more fulfilling work and a path to career progress. More quotes mean growth, and space to ask: “What could be even better?” Give people the keys and the autonomy to find ways to improve and innovate.

AI fluency: you can’t trust what you don’t understand

Successful technology change starts with business strategy. I know that sounds obvious, but it’s easier said than done. Hearing “the big why” from leadership can ease fears and boost engagement from the start.

If people don’t know enough about a tool to use or trust it, we’ve got a big problem. The CII calls this the “AI fluency gap”: firms deploying AI faster than their people can get to grips with it, let alone challenge it.

We found that AI talent shortage is one of the top barriers to scaling AI (34%), and 39% of insurers cite a lack of in-house data skills. People who don’t understand AI tend to either avoid it or misuse it, leading to more shadow IT to keep compliance and security teams up at night.

Remember the early days of satnav? There were many stories of people driving into lakes, or towards cliff edges, because the satnav told them to. We must guard against automation bias and complacency: relying on technology so much that it impairs our judgement.

Our teams need the training and confidence to use AI well, and to push back when it’s wrong. For many, this is a learned skill that takes time and structure before it feels comfortable.

Let the people build

We have a pensions client (a different sector, but the same challenge) that’s training its people to build AI agents. Here we’re seeing how autonomy builds trust. People can solve their own problems from their expert perspective.

Instead of waiting for a programme that may or may not deliver the technology they need, they can change things directly. They can self-heal, within the right boundaries.

You may already have the vehicle for it. Most insurers run apprenticeship schemes, so you could offer an AI apprenticeship alongside business management. It can also be funded: since April 2026, levy funds can pay for short AI courses covering topics like AI adoption, governance and transformation, as well as full apprenticeships.

Fluency should also bring capacity. The more grunt work people can offload to automation, the more time they have to help shape bigger changes.

Technology adoption: build it with them

I often hear that big bang programmes are dead. But they’re still common, because they make sense to the purse-string holders. They don’t tend to be technologists; they work to the rhythm of budget cycles.

A fixed eight-figure sum over three years is a known cost you can plan for. But it can lead firms to sign up to the whole thing before anyone fully understands the problem. As the costs creep up, the scope shrinks. You start with the whole and end up chipping away at its potential. It’s hit and hope. And we’re not natural gamblers in this industry…

Adoption, naturally, suffers. Without close coordination with the users who know the work best, the features that would have made the tool useful get de-scoped. Unfortunately, those users are tied up in BAU, so they can’t help shape it, and it’s built without them. When it lands, the response is more often than not: “What the hell is this?” The current system may be imperfect, but it still does the job. So they don’t use the new tool, or they find a workaround.

Start where there is trust

Storytelling is key. We need wins that inspire, showing practice over theory and reality over words. Pick the path of least resistance, where there’s already trust. Our report found that 51% of insurers already use AI for risk modelling and stress testing, where the data is typically high quality and reliable. That’s an ideal place to start: work with the people who will use the AI solution, run small tests, measure them and validate what works.

Three guiding principles

So what can you practically do next? Whatever you do, it helps to remember these three principles:

Culture: tell people upfront where the time saved will go, and make sure it doesn’t just get absorbed back into BAU.
Fluency: launch an AI cohort through your apprenticeship scheme. Track progress, formalise feedback, and make sure it’s an enjoyable and worthwhile experience.
Adoption: pick one use case tied to a business objective, and build it with the people who know the work best.

You’ll notice that none of this is about technology. It’s about leadership, communication and bringing the organisation together. It will take time, naturally, but it’s time better spent than on multi-year programmes that waste heaps of money and repeat history.

Assess your AI maturity

Our AI Maturity Assessment compares your organisation with peers across the insurance sector, including composite insurers, specialty insurers, MGAs and brokers. It finds your biggest blocker and identifies where to focus next.

Take the 3-minute assessment