The AI readiness problem is not the technology. It is the sequence.

Insurance leaders have more access to capable AI tools than at any prior point in the industry’s history. The tools are more powerful, more accessible, and, in many cases, more directly applicable to insurance workflows than the technology of previous adoption cycles. And yet, the rate of meaningful production deployment remains low relative to the volume of pilots and vendor conversations. The problem is not the technology. It is the sequence in which organizations are approaching adoption.

Kirstin Marr, founder and CEO of Lead The Machine, offers a framework developed from 25 years of technology and insurance leadership that cuts through the adoption conversation with unusual clarity: start with the business problem, get the data foundation in order, bring people along, and measure outcomes precisely. This is not a new principle. What is new is the cost of ignoring it.

Starting with the tool is the most common and most expensive mistake

Insurance organizations that begin AI adoption by selecting a tool, platform, or vendor and then searching for problems to apply it to are structuring the process backwards. This approach generates pilots that cannot scale because they are not anchored to specific operational challenges that matter financially or strategically. It also generates skepticism in the workforce when tools are introduced without clear connection to the work people are actually doing.

The concrete insurance problems that AI addresses most effectively include risk assessment in underwriting, customer operations efficiency, and workflow automation for repetitive processing tasks. But the specificity matters. “Improve underwriting” is not a business problem. “Reduce the time underwriters spend processing documents to extract complex risk factors” is. The difference between these formulations determines whether AI deployment produces measurable results or generates activity without outcomes.

Data is the prerequisite that organizations are still skipping

Marr’s emphasis on data quality, accessibility, and governance as foundational conditions for AI success echoes what practitioners across the industry consistently report but organizations consistently underinvest in. The sequencing failure here is almost universal: organizations want to run AI on their data before their data is in a condition to support it.

Data that is scattered across legacy systems, inconsistently formatted, and lacking governance infrastructure will not produce reliable AI outputs. It will produce outputs that are occasionally useful in pilots and systematically unreliable in production. The investment in getting the data foundation right is expensive and time-consuming. It is also the investment that makes every subsequent AI application more valuable. Organizations that skip this step are not moving faster. They are building on a foundation that will require remediation later at greater cost.

People are the adoption layer that technology budgets ignore

The third element in Marr’s framework, bringing people along, sounds obvious and is routinely under-resourced. Change management in technology deployment is typically treated as a communication exercise: announce the tool, provide training, monitor adoption metrics. What actually determines whether AI changes how work gets done is whether employees understand how the tool connects to their specific role and whether their incentives are aligned with adopting it.

Marr’s observation that customers still want to buy from people is a useful grounding point for insurance specifically. AI in sales and distribution should be structured to augment human relationship capacity, handling the prospecting, data retrieval, and preparation work that currently consumes time producers could spend with clients. Organizations that deploy AI as a substitute for human judgment in relationship-intensive roles will find that the efficiency gains do not translate to revenue outcomes.

A ReSource Pro perspective

The organizations succeeding with AI adoption in insurance are following a sequence, not just deploying tools. They identify specific operational problems, prepare the data infrastructure that allows AI to run reliably on real production data, equip the people whose workflows are changing, and measure outcomes precisely enough to know whether the investment is working. Each of those steps depends on organizational discipline that precedes the technology conversation. The combination of process maturity, data governance, and workforce readiness is what separates organizations that scale AI from those that accumulate pilots.

Change is slow until it’s not: a practical roadmap for AI implementation in insurance
Author: Kirstin Marr, Founder and CEO, Lead The Machine
Publication: The Insurance Lead
Original Publication Date: April 28, 2026