When AI Begins Running the Insurance Workflow
In this series, I’ve explored the Platform Insurance Company: an operating model built on unified technology platforms, embedded intelligence, micro-vertical expertise, and partner-driven innovation. Now I want to get specific about what this looks like in practice.
Forget the productivity gains. Forget the checkbox mentality about AI. Autonomous agentic operations mark the point where AI systems do not simply analyse data or recommend action. Working seamlessly across an ecosystem of disparate platforms, they access the full operational environment, which allows them to coordinate processes across underwriting, claims, customer engagement, and risk management towards a complete and radical transformation of how we serve customers.
Tangible Results, Not Hype
For claims, the agentic workflow delivers faster claims resolution and improved customer experience. Instead of multiple manual steps, the entire process of submit, review, assess, and approve is dynamically orchestrated by multiple AI agents swarming to solve complex cases quickly and easily. They analyse claim documentation and images, assess damage using computer vision, check historical patterns for fraud indicators, retrieve policy details, estimate repair costs, and initiate settlement recommendations.
What human teams once took days, weeks and months to complete now takes seconds, minutes and hours at most. Agentic seeks to eliminate process inefficiencies, overpayments, and missed subrogation, which drain 5-10% of total claims cost, resulting in avoidable material losses. It also tackles bottlenecks which often unnecessarily delay claims for weeks and months and cost companies millions in lost opportunities and inaccurate filings. Underwriting is another area where dramatic changes are happening.
Where human underwriters used to spend 40% of their time on administrative tasks, AI agents now continuously analyse risk signals from historical claims data, environmental risk models, economic indicators, behavioral data, IoT and telematics inputs.
In terms of customers, relationships have traditionally been reactive. Policyholders contact insurers when something goes wrong or when policies need to be renewed. Customers, who are spammed in all other areas of their life, are aware of the transactional nature of such communications and act (or don’t act) accordingly.
In the new operational model, agentic AI enables a proactive, fast and affirmative way of doing business. AI agents are configured to monitor behavioral signals, life events, and risk indicators to anticipate customer needs. When AI approaches customers who might require additional coverage, or at risk they might not be aware of, or are going through life changes that might require policy adjustments, they are viewed as more human and empathetic, according to a survey by Lemonade.
With personalised communication across digital channels, insurance relationships become continuous rather than episodic, which makes a profound change in customer engagement.
I am not talking about simple automation or process efficiency. I mean that AI is doing something that human beings cannot: see across the entire organisation, analysing millions of data points and making decisions based on that data. There is nothing speculative in these results, nor do I intend to induce any of the usual paranoia about AI.
What Agentic AI Actually Means
Agentic AI refers to intelligent systems that understand goals and objectives, analyse complex operational data, execute actions across multiple systems, and adapt decisions based on new information.
Instead of waiting for humans to initiate every step in a workflow, AI agents coordinate activities dynamically to achieve a defined business outcome. In the insurance context, this shift means AI systems manage large portions of the operational lifecycle, not by replacing human expertise, but by orchestrating the workflow in which that expertise operates.
This point cannot be overemphasized. Against recent advertisements presenting AI as a replacement for people, AI and human aren’t opposites — they’re interdependent. The underwriter’s expertise becomes the training signal. Without human judgment, strategy, and relationships, agentic operations don’t function. Without AI, human expertise can’t scale.
As I discussed recently in an interview with Life Insurance International, human capital (judgment, relationships, ingenuity) and token capital (the AI capability the firm owns and controls) are the same. Human beings spend token capital, and token capital relies on human beings. Without them, agentic operations depend on humans for complex judgment, strategy, and relationships. The underwriter’s expertise becomes the training signal.
Insurance Reaches that “Netflix Moment”
Every major industry eventually experiences a moment when technology stops simply supporting operations and begins actively driving them. We have seen retail experience this shift with digital commerce platforms. Finance experienced it with algorithmic trading systems. Now it’s insurance’s turn – that “Netflix moment” when the industry either completely transforms its business and dominates the future or goes the way of Blockbuster.
While many insurance companies are still deeply invested in the previous generation of automation for document processing, routing claims to adjusters, triggering underwriting reviews and processing payments and billing events, there are many insurers using AI agents to coordinate workflows, platforms expand capabilities, and operations continuously optimize for measurable outcomes.
Agentic AI can only function effectively within unified platform environments. If underwriting systems, claims systems, customer platforms, and data environments remain fragmented, AI agents cannot coordinate workflows across the enterprise, which is why the shift toward platform architecture is so critical.
The insurers already operating this way aren’t running pilot programmes. They’re running the business. The question isn’t whether agentic AI will define insurance operations, it’s whether your organisation will define it or react to it.
Read the other blogs in the series:
5 Insurtech Capabilities the C-Suite is Discussing: Insider Perpective
How a Platform Approach Drives Autonomous Insurance Outcomes
Outcome-Seeking AI: Insurance’s Next Platform Evolution
The Case for Building Insurance Platforms That Actually Understand Insurance
