Steve Walker and Camelot Consultants published a challenge and opportunity- ‘Strategic Claims Management: Taking Claims From an Operational Necessity to a Strategic Asset’ which is well worth a read. Just follow the link to see why the author stated "
“The effective management of claims is one of the most central capabilities in the insurance value chain. It is the “moment of truth” – where brand promise becomes a lived customer experience. Whilst the most successful insurers and MGA’s recognise it as a core capability, it is not always fully leveraged as a source of competitive advantage.
Strengthening claims capabilities across all activities is costly - and not all investments will drive equal returns. The key lies in identifying which levers to pull and where you can move the needle to align improvements with the broader strategic priorities of the business".
What about the practical details of how to leverage AIs across claims? Chris Brown and I decided to publish a series of articles taking you through all aspects devoid of hype and full of hands-on experience. The first article is below, and you'll find details of the full series at the end.
As most in the insurance industry know, experienced adjusters and claims handlers are retiring faster than they're being replaced. This isn't a future problem. It's happening now. This is one of the motivations to adopt AI. To take the admin burden off claims handlers and leave them more time to deal with value-added tasks. Not least, be available when customers want to talk to a person rather than a bot.
I've spent tdecades working with transformation, a decade on claims whilst Chris has buit AI systems for insurance claims. I've seen what is possible, what we got working, what doesn't, and what gets quietly shelved when production realities collide with demo promises.
What I've observed is that the technology doesn't fail. It's that the deployment context is harder than anyone admits. And the pressures driving adoption are real enough that insurers can't afford to get this wrong. Let my experience help you avoid that calamity.
Below you’ll find the following contents
The talent crisis is real
Customer expectations have shifted
Demographic shifts change everything
Cost pressures are relentless
The seductive promise of agentic AI
Insurance is not a typical enterprise deployment
Future articles offer practical help
Let me start with why the pressure exists.
The Talent Crisis Is Real
The numbers are stark. According to the Chartered Insurance Institute (CII), 25% of the UK insurance sector will retire within the next decade [1]. The London Market Group's 2024 research found that the number of professionals aged over 50 now equals those under 30. That's a demographic imbalance years in the making [2]. RSA Insurance data shows 26% of UK insurance employees are already over 50 [3].
The pipeline isn't filling either. CII research found that only 4% of young people report wanting a career in insurance [3]. Workplace attrition reached 18% at some leading UK firms in 2023, with more than a quarter of the workforce leaving their positions in the last three years [4].
Meanwhile, the skills gap is widening. The CII's 2024 Insurance Talent Report found 72% of insurers are struggling to find talent with the right technical and data skills, and nearly 60% worry they'll lose staff to other industries [5]. The Davies People Report found 63% of senior managers acknowledge a digital skills gap in their teams, with 30% describing it as "very serious"[6].
The US faces similar pressures. The Bureau of Labor Statistics projects that while the total number of claims professionals will decline by 5%, the industry will face approximately 21,500 job vacancies each year over the next decade [9]. The US Chamber of Commerce warns that over the next 15 years, about 50% of the current insurance workforce will retire, leaving more than 400,000 positions unfilled [10].
The institutional knowledge leaving the organisation represents decades of accumulated judgment. How to interpret ambiguous policy wording. When a claim "feels wrong." How to handle a distressed customer. What questions to ask that aren't on any script. This expertise was built over thousands of claims, years of mentorship, and countless edge cases that never made it into the training manual.
You can't replace that with a two-week onboarding programme. How do you cope with demanding customer expectations.
Customer Expectations Have Permanently Shifted
People want to report claims at any time, even if it is midnight, from their phone. They want immediate acknowledgement that the claim has been lodged. They want to track progress like a delivery. They want resolution in days, not weeks.
The data confirms this shift.
Insurity's 2025 Digital Experience Index found 64% of consumers would consider switching insurers for a better digital experience [7]. Mobile is now the default. 86% of users access their policy on a mobile device, and over 80% prefer to manage it on their phone [8].
But here's the nuance. Only 15% want a fully digital, self-service experience. Nearly half (48%) prefer a digital-first approach with the option to speak with someone when needed [7]. Customers want convenience, speed, and 24/7 access, but they also want human support available for complex or stressful situations claims.
They're comparing their claims experience to their banking app, not to what insurance was like a decade ago. The requirement to phone someone during business hours feels archaic to anyone under 40. The expectation that they'll need to repeat information multiple times appears to be a sign of incompetence.
This isn't unreasonable. Other industries have transformed their customer experience. Insurance, largely, hasn't. The gap between expectation and reality grows wider each year. At the same time, the thin margins insurers work with means that cost pressures are challenging.
The Demographic Shift Compounds Everything
While customers expect more from claims teams, the talent crisis makes it more difficult to deliver.
The handlers who can be recruited often lack the deep experience of those leaving. Training takes years. Complexity isn't reducing. If anything, new risks like cyber, climate, and supply chain disruption are adding complexity that even experienced handlers find challenging.
You're trying to handle more claims, with less experienced staff, while customer expectations rise and margins compress. Something must give. Unless agentic AI promises are real.
Cost Pressures Are Relentless
Combined ratios leave limited room for operational inefficiency. Every manual touchpoint that could be automated represents a margin that competitors might capture. Every experienced handler doing work that could be systematised is a capacity that could be redeployed to complex cases that genuinely need human judgment.
The pressure isn't just "do more with less." It's "transform fundamentally or watch margins erode." Incremental efficiency gains won't close the gap. BUT- cost reduction and customer experience are often conflicting goals. Demographic trends sharpen the conflict.
The Seductive Promise of Agents
Into this perfect storm comes agentic AI. Systems that don't just answer questions but take autonomous actions.
The pitch is compelling. Let AI handle routine triage. Let it assess evidence. Let it make straightforward coverage decisions. Let it draft the correspondence. Let it, eventually, process straightforward claims end-to-end.
Free your experienced handlers from repetitive work. Let them focus on complex cases that genuinely need human judgment. Scale capacity without scaling headcount. Meet customer expectations for speed and availability without breaking the cost model.
The technology works. I've built and seen impressive demonstrations. Natural language understanding that accurately captures claim details. Document extraction that pulls the right information from photos and reports. Coverage analysis that identifies relevant policy terms. Fraud pattern detection that flags anomalies.
The appeal is obvious. The pressure to adopt is real. But insurance is a particularly challenging deployment arena.
Insurance Isn't a Typical Deployment Context
Here's what the demos don't address. Insurance claims aren't a productivity optimisation problem. They're a trust, compliance, and evidence problem.
When a coverage decision is challenged, you need to explain exactly how it was reached. When a regulator asks how customer outcomes were determined, you know "the AI decided" isn't an acceptable answer. When a claim goes to litigation five years from now, you need to produce the decision audit trail and defend it. Right down to conversations held at the time.
Insurance has spent decades developing structured workflows, handler scripts, training programmes, and quality assurance processes precisely because these scenarios recur frequently. The question isn't whether AI can process a claim. It's whether you can defend that processing to regulators, courts, and customers. Potentially years after the decision was made.
The vendors demonstrating will not be in the room when the FCA asks questions. You will be.
Future articles in this series will show you why, when and where to plan, build and deploy agentic AI and not be left alone to defend decisions made in any claim.
What This Series Will Cover
This is the first in a series examining what is likely required to deploy agentic AI responsibly in insurance claims.
Not the vendor pitch. Not the sceptic's dismissal. A practitioner's reality check.
We'll cover:
The claims spectrum: why treating "claims" as monolithic is the first mistake, and how to map your portfolio to appropriate AI approaches
When nobody can agree on the definition, the governance implications depend entirely on which type of system you've actually built
The production gap: why impressive demos sometimes don't translate to production deployment in regulated industries
The automation trap: why "set and forget" automation will fail, and what sustainable deployment requires
The measurement problem: the double standard between human and AI accuracy, and what mature measurement looks like
A realistic roadmap: practical guidance on what to deploy where, and the governance infrastructure you need
Agentic AI offers a genuine response to the pressures insurance faces. But capturing that opportunity requires understanding what's different about deploying AI in a regulated, evidence-dependent, litigation-exposed industry.
The insurers who get this right will build sustainable advantage. Those who pursue vendor demonstrations without building the foundations will find themselves explaining failures to regulators, customers, and boards.
The stakes are too high for wishful thinking. Let's get specific about what works.
Please feel free to contact me anytime to answer your specific questions
Next in series: The Claims Spectrum - why not all claims are equal, and what that means for your AI strategy.
Authors: Chris Brown - The Build Paradox , Mike Daly - Insurtech World
References
[2]: London Market Group, "London Matters" data (May 2024).
[4]: Burton Recruitment, "The Changing Face of Insurance Industry Recruitment" (January 2024).
[6]: Davies Talent Solutions, "The top recruiting challenges for insurance firms in 2025" (June 2025).
As AI automation becomes increasingly prevalent in the insurance industry, transparency alone is not enough; responsible governance and expert oversight are also essential. Insurers that consistently demonstrate the responsible use of AI, disclose model use, validate fairness, and communicate openly will distinguish themselves in the marketplace. By combining forensic accounting, digital forensics, engineering, and behavioral science, insurers can gain a deeper understanding of the technical and ethical aspects of AI adoption. This not only strengthens internal decision-making but also enhances auditability.
unknownx500
