The problem of the missing ladder
The insurance sector is caught in a severe demographic squeeze at both ends of the career ladder. According to an article published by RSM International, over a quarter of the UK insurance workforce is over 50, and half could retire within fifteen years, taking decades of accumulated knowledge with them. At the same time, graduate vacancies fell by 18 per cent in 2025, over half of brokers struggle to recruit people under 30, and only 4 per cent of young people view insurance as an appealing career.
Over the last year, I have learnt that almost all insurers have automated workflows handling routine paperwork, and yet they still face the problem of replacing the experience of an aging workforce. Automation is great for standard patterns, but it struggles with grey areas: ambiguous policy wording, non-standard risks, or claims that simply look wrong. Spotting these issues takes real judgment, and judgment only comes from years of seeing how things play out in practice.
By automating basic tasks without replacing the learning process, the industry has accidentally pulled away the ladder that leads to experience. The senior experts who catch million-pound mistakes will retire within a decade, and the junior staff coming up behind them will not have had enough exposure to take over.
The solution: reimagining the career ladder for the age of automation
Having worked closely with insurers on their digital transformation journey, I believe we need to tackle this challenge at three connected levels. Each level addresses a particular limitation and can be tackled independently, and the good news is that AI is part of the solution in each case.
Level 1: Practical Knowledge Preservation so decades of corporate memory do not walk out the door when senior staff retire.
Level 2: Accelerating learning through simulation to replace the hands-on exposure that basic automation has taken away.
Level 3: Guided real-world decision-making where senior experts act as guardrails, helping junior staff build genuine judgement rather than just following software prompts.
Level 1: Practical Knowledge Preservation
Most of the knowledge an insurer needs already exists in old policies, guidelines, past claims, and pricing sheets. You do not need a massive, costly new project to capture it. But you do need to make sure that you can find it. The priority is to make existing information easy to retrieve through a simple search interface – before senior experts retire.
One of the major obstacles to success I have seen is jargon. Different departments often use completely different words for the exact same thing. What underwriters call a facility, the distribution team might call a scheme. Standard search tools break down because a junior handler searching in their own terms will miss documents written by another team.
Trying to force every department onto a single corporate glossary is a long, painful exercise that I have yet to see succeed. A far smarter approach is using semantic hybrid search. This pairs traditional keyword matching with context search so the system understands intent. A junior can ask a question using their team’s everyday vocabulary and get relevant answers from across the entire business, regardless of the jargon.
Level 2: Accelerating Learning Through Simulation
Decades of closed claims, declined risks, and past policy files offer an ideal foundation for practical learning. Turning these historical cases into a practice simulator gives junior staff a safe environment to experience real scenarios and build judgment. AI makes this kind of bespoke, personalised and company-specific training easily achievable.
I believe this kind of training is best implemented in phases starting with a single line of business. Take a few hundred closed files where the outcome is already known, remove any personal data, and set up a pilot with two or three trainees.
Junior staff can work through these cases and make their own underwriting or claims decisions. Because the system checks their work against what actually happened, the feedback is immediate and trustworthy. Progressively, the simulator can introduce increasingly complex edge cases, helping trainees gain years of exposure in a matter of months. Repeat across all lines of business to cover your entire organisation.
Level 3: Guided real-world decision-making
There is an understandable concern in the market that automating routine tasks will deprive junior staff of basic practice. However, if trainees remain in the loop, actively reviewing automated casework and making their own decisions on underwriting or claims, AI can help them see a much wider variety of scenarios than traditional manual work allows.
But the way you do this really matters. If trainees see the system’s answer straight away, it is too easy and dangerous to just nod along without any critical thinking and agree with the AI draft. All AI models come with inherent bias. Blindly accepting their recommendations can cause serious damage to the business, for example, a claim denied due to ethnicity.
The idea is that the AI performs the routine paperwork and recommends next actions. The juniors review the routine work, form an opinion and a proposed decision before looking at the AI recommendations. The AI recommendations are to be used only as a benchmark to test their decision. If a junior is still unsure how to proceed after reviewing related previous claims, they can then reach out to senior staff for guidance.
The time to act is now
Fixing the missing ladder is ultimately not about replacing people with technology. It is about using AI to pass the torch to the next generation before it is too late. By preserving existing knowledge, simulating real cases, and guiding live decisions, insurers can build genuine judgment in junior staff far quicker than traditional pathways allowed.
The reason to act now is simple: use the insight and oversight of the senior experts who are still in the business today before the ladder’s gone for good.
References
RSM International: https://www.rsmuk.com/insights/advisory/the-future-of-insurance-an-ageing-workforce-and-a-growing-talent-gap
