The Quiet Revolution Inside Insurance: Why AI in Workflows Is No Longer Optional

The industry has always prided itself on prudence. But the gap between carriers who embed AI into their daily operations and those still running on manual workflows is widening fast — and quietly.

There is a well-worn rhythm to how insurance gets done. A submission arrives. Someone reads it. Someone else verifies it. A decision gets documented, routed, approved, and filed. At each stage, the same data passes through different hands, often re-entered, re-checked, and occasionally lost to inbox overload.

That rhythm hasn't changed much in forty years. Until now.

Across the industry, something is shifting — not in the loud, disruptive way that fintech promised a decade ago, but methodically, inside the operational core of carriers who have quietly started replacing manual handoffs with intelligent workflows. The results are difficult to argue with.

The Gap Between Pilots and Production

Most insurers today are aware of AI. According to a 2025 Conning survey, nine in ten are actively evaluating or implementing it. But awareness and transformation are very different things. The same data shows only about one in ten has achieved scaled deployment in any single function.

This is the pilot trap. Organizations run proof-of-concept projects, see promising results, and then stall — caught between technology enthusiasm and the hard work of rewiring how work actually flows. McKinsey, in their detailed 2025 analysis of AI in financial services, was direct about what separates the leaders from the followers: it isn't the sophistication of the models. It's the decision to stop tinkering at the edges and embed AI into the entire operating model of a domain.

To create lasting business value from AI, insurers need to set a bold, enterprise-wide vision for AI's potential, and deeply, fundamentally rewire how they operate.

McKinsey & Company — The Future of AI in the Insurance Industry, 2025

The carriers who have crossed this threshold are compounding their advantage every quarter. And unlike previous waves of technology investment, the costs of not moving are becoming structurally embedded — in combined ratios, in talent acquisition friction, in the growing expectations of commercial policyholders who now compare their insurer to every other digital experience in their life.

Where the Friction Actually Lives

To understand why AI in workflows matters, it helps to map where time actually disappears in an insurance operation. It rarely vanishes in a single dramatic bottleneck. It bleeds out across dozens of small, routine tasks that individually seem manageable — and collectively consume enormous capacity.

A junior underwriter rekeying submission data from an email into a policy system. An actuary waiting three days for a BPO team to compile claim statistics that should take three hours. A claims handler manually reviewing documents to verify coverage before informing a policyholder. A compliance officer reading through regulatory changes and manually flagging which product lines are affected.

Each of these is, at its core, a structured reading and categorization task. And that is precisely where AI is most capable today.

These aren't speculative use cases. Claims processing already leads AI adoption across the industry at 64%, with fraud detection close behind. Underwriting, currently at around 14% adoption, is projected to reach 70% by 2028 — making it the fastest-growing application in the near term.

What Deloitte's 2025 Assessment Reveals

Deloitte's Digital Insurance Maturity report — one of the most comprehensive assessments of its kind, covering 93 insurers across 16 EMEA markets — makes a finding that the industry should sit with for a moment: the shift toward digitally enabled engagement is no longer optional, it is a core business imperative.

The report is particularly pointed about where the gap opens up. Leading insurers are moving beyond transactional portals to design omnichannel journeys powered by real-time analytics, conversational AI, and behavioural intelligence. They aren't doing this because they love technology. They're doing it because customers now expect 24/7 self-service, instant claims status, and contextual communication.

But Deloitte's most important observation is about the difference between surface-level and structural change. A polished claims portal built on top of a fragmented back-end is a cosmetic improvement, not a transformation. Genuine digital maturity requires modernising policy administration systems, claims engines, and data architecture to enable real-time responsiveness throughout.

The Revenue Case, Not Just the Cost Case

The conversation around AI in insurance often starts with efficiency — processing times, headcount, cost reduction. These are real benefits, but they undersell the opportunity significantly.

McKinsey's February 2026 analysis, reported by Reinsurance News, estimates that generative AI could unlock $50 to $70 billion in new insurance revenue. The mechanism isn't automation displacing people. It's AI enabling insurance to go places it couldn't go before — hyper-personalised products priced with better precision, coverage offered at the moment of need through embedded distribution, and risk assessment that incorporates data sources that were previously too unstructured to use.

Underwriting, in particular, is experiencing a quiet revolution in accuracy. AI-assisted risk models that analyse complex, multi-variable datasets are improving underwriting accuracy by over 50% in some implementations. Better accuracy means better pricing. Better pricing means lower combined ratios and the confidence to write more business.

The revenue uplift from AI isn't a side effect of efficiency. It's a direct consequence of making better decisions, faster, at scale.

The Culture Piece Nobody Talks About

Deloitte's report includes an observation that stands apart from the technical analysis: AI's impact on insurance will be as much about people and culture as it is about algorithms.

This is important, and often underweighted. Research cited in the report found that employees perceive their employer as significantly less empathetic when AI tools are introduced without adequate context or support. The technology question is the easier half of the problem. The harder half is building an environment where people feel equipped and motivated to work alongside intelligent systems — not threatened by them.

A 2025 Counterpart survey of young insurance professionals found that 69% see AI as a collaborator rather than competition. That's a favourable starting point. But the same survey identified slow technology adoption as the industry's biggest hurdle — not from employees resisting change, but from organisations not moving fast enough to meet the moment.

The carriers who get this right won't be the ones who deployed the most sophisticated models. They'll be the ones who designed workflows where an underwriter with AI assistance writes more business, handles more complex risks, and builds stronger broker relationships — and understands exactly why that's better for everyone.

From Pilots to Production: A Practical Frame

If most insurers are stuck in pilot mode, what does the path forward actually look like? McKinsey's guidance is instructive: rather than fragmenting AI investment across dozens of disconnected use cases, the most effective approach is to pick one to three domains and transform them end-to-end. Claims is often the right starting point — it has the highest existing adoption, the clearest ROI benchmarks, and the most immediate impact on policyholder experience.

From there, the logic of integration compounds. A claims system that AI has optimised generates better data. Better data improves the pricing models that feed underwriting. Better underwriting improves risk selection. Better risk selection improves the claims experience. Each domain feeds the next.

The organisations that understand this aren't asking whether to invest in AI. They're asking which domains to transform first, and how to build the data and technology foundations that will make each subsequent step faster.

Full AI adoption in insurance jumped from 8% to 34% year-over-year between 2024 and 2025. The industry is not moving slowly. The question is which side of that gap you're on.

Datagrid Research — AI Agent Statistics in Insurance, 2025

The Competitive Reality

Insurance has historically competed on relationships, brand, and the size of its balance sheet. Those advantages haven't disappeared. But they're being supplemented — and in some segments, supplanted — by operational capability.

A carrier that can quote faster, pay claims faster, detect fraud earlier, and serve policyholders around the clock doesn't need to win on price alone. It wins on the quality of the experience and the reliability of the outcome. In commercial lines, where brokers place business with carriers who make their lives easier, this is already a differentiator. In personal lines, where aggregators have commoditised price comparison, it may be the only differentiator left.

The window for a measured, deliberate approach to AI adoption is narrowing. Not because the technology is running away from anyone, but because the organisations that started earlier are accumulating compounding advantages in data quality, model accuracy, and operational fluency. The gap between them and those who are still piloting is widening every quarter.

The good news is that the patterns for doing this well are now well established. The workflows are understood. The ROI benchmarks exist. The principal question facing most carriers today isn't whether AI belongs in their operations. It's whether they have the right partner to help them build the systems that will actually run in production.