Skip to content
← Kathas
Conversation · 3 min read

Not all AI is built for insurance

Most “AI in insurance” is still task automation. The next wave is about supporting decisions, and it has to be built for the industry from the ground up.

I
Words by
InsureStory

Walk into any insurance conference or strategy meeting and AI is everywhere. Underwriting, claims, distribution, service: the message is that AI is transforming insurance. That’s true. But most of what we call AI in insurance today is still task automation. Not judgment. Not real insurance intelligence.

The last decade brought wave after wave: robotic process automation, predictive analytics, chatbots, workflow engines and AI copilots. Each made work move faster. Few helped anyone understand the work itself. Extracting data from documents and moving submissions between systems is valuable, but limited.

Insurance is not just about processing information. It’s about making decisions under uncertainty.

Should we write this risk? At what premium? On what terms? Does it comply? Those are judgment calls, not checklist tasks. That is the idea behind decision intelligence, the approach taken by Perpendo AI: interpret complex submissions rather than just extract data, support underwriting judgment, produce audit-ready decisions and build compliance into every step.

Context is the hard part. General-purpose models are broadly trained and industry-agnostic, while insurance runs on layered risk, state-by-state regulation and highly specific policy structures. Perpendo frames it around five Rs: risk, rules, relationships, regulations and responsibilities. You can’t bolt that knowledge on afterwards. If an AI system doesn’t understand the domain, it doesn’t just fail. It creates risk.

Compliance is the other dividing line. AI adoption in US insurance isn’t held back by technology so much as by regulation, audit requirements and the need to explain decisions. Treating compliance as a later add-on doesn’t work. If you can’t explain a decision, you can’t defend it, and if you can’t defend it, you shouldn’t automate it.

For carriers and MGAs, the payoff is faster underwriting, more straight-through processing, more consistent decisions and better audit readiness. The bigger change is human. Underwriters spend less time on repetitive tasks and more on complex risks and broker relationships. AI handles the noise; people handle what matters.

The first phase of AI in insurance was about digitisation and efficiency. The next will be about understanding risk. The winners won’t be the ones who automate the most, but the ones who understand insurance best.

終

More Kathas

Conversation · 3 min read

Are brokers becoming tech companies?

AI-native underwriting, digital placement and predictive dashboards are changing how brokers create value, without changing why clients need them.

An InsureStory perspective
Words by InsureStory