Abhishek Kumar is an AI product leader at NY Life.
The life insurance application process has not changed much in 50 years. Applicants fill out forms, schedule medical exams and wait. Weeks pass before anyone reviews the file. For healthy people in their 30s and 40s, the delay often feels excessive.
Artificial intelligence is beginning to change this. I lead AI product initiatives at New York Life, building generative AI and machine learning solutions across underwriting, marketing and service operations. In my work across financial services and insurance, I have seen AI tools reduce underwriting timelines from weeks to hours by accelerating how information is gathered and reviewed. Technology does not replace human judgment. It supports underwriters by reducing manual work and surfacing relevant information faster.
The barrier to enterprise AI adoption is rarely the model itself. The harder problem is operational. Regulated institutions need systems that fit existing workflows, satisfy compliance requirements and earn employee trust. That gap between technical capability and institutional deployment is where many AI initiatives succeed or fail.
The Problem With Traditional Underwriting
Traditional underwriting relies on the manual review of medical records, lab results, prescription histories and applicant disclosures. In my experience, the process typically takes three to six weeks, though timelines often stretch longer for applicants with complex medical histories or high coverage amounts.
This creates friction throughout the process. Applicants abandon applications because of delays, agents lose momentum and operations teams absorb high manual processing costs.
According to McKinsey, insurers deploying AI across underwriting and claims domains are seeing measurable gains, with one leading carrier “cutting liability assessment time for complex cases by 23 days” and “reducing customer complaints by 65 percent” through AI-driven automation alone. Those efficiency gains translate directly to faster turnaround times and lower per-application costs.
How AI Fits Into The Workflow
AI models are most effective when they handle specific tasks within underwriting. They can extract data from documents, organize medical records into structured formats and flag anomalies or risk patterns for underwriter review.
The goal is not to replace underwriters. It is to remove repetitive work, so experienced professionals can focus on cases that require judgment. Straightforward cases may move through automated review, while complex files still require human evaluation.
In my experience, most AI underwriting failures are not technical. A team may build a model that performs well in a demo but struggles to reach production. In other cases, the system may launch successfully, but nobody uses it because the workflow was designed around what the model could do, rather than how underwriters actually work.
An underwriter does not need more information. They need the right information surfaced at the right point in the workflow without leaving the system they have spent years mastering. Getting that sequencing right is harder than building the model itself.
Building AI Products For Regulated Industries
Life insurance operates under strict regulatory oversight. Every underwriting decision must be explainable and documented for regulators. AI systems used in this environment require strong governance, including testing before deployment and ongoing monitoring after launch.
Fairness is another major concern. AI systems can unintentionally learn patterns from historical data that disadvantage certain groups. The NAIC has issued guidance that requires insurers to implement bias testing and model validation processes for AI tools used in underwriting.
Underwriting systems also handle highly sensitive medical and financial information, requiring strict security standards.
These realities shape how product teams work. Data scientists, engineers, compliance officers and underwriters all need to contribute to development. Many organizations struggle because these groups operate in silos and define success differently.
How To Embrace AI
For leaders navigating this transition, I’ve found that several principles consistently matter.
First, compliance should be treated as a design input rather than a review gate. Organizations that involve compliance teams early, in my experience, spend far less time retrofitting systems later.
Second, teams need a shared definition of failure before development begins. Data scientists often define failure as model error. Underwriters define it as poor decision-making. Compliance officers define it as a decision that cannot be audited.
Third, guardrail infrastructure must be built before deploying AI systems into production. In regulated industries, the question is never whether guardrails are necessary. The real question is whether they are integrated into the platform from the beginning or added after a failure.
Finally, organizations should begin with the highest-volume, lowest-complexity workflows. Your goal should be to build operational trust and institutional confidence.
The carriers that succeed will not necessarily be the ones with the most advanced models. Instead, those carriers must understand their workflows deeply enough to know where intelligence actually improves outcomes.
The Business Case For AI Underwriting
The economic case for AI underwriting is becoming increasingly clear. Industry research from Datos Insights indicates that carriers actively deploying AI are seeing “underwriting accuracy improvements of 15% to 45%” alongside claims processing speeds that are 30% to 50% faster while also improving decision consistency.
I’ve found that those gains improve both operational efficiency and customer experience. Carriers that automate routine underwriting tasks can redirect experienced underwriters to higher-value work, all while scaling operations without proportional increases in headcount.
This shift is becoming more important as customer expectations continue to evolve. Consumers increasingly expect the same speed and convenience from insurers that they already receive from digital banking platforms and retail services.
What Comes Next
The broader direction of the industry is already clear. Carriers that successfully integrate AI into underwriting workflows may be able to offer faster decisions, see lower operating costs and deliver more consistent customer experiences. Those that delay adoption risk higher costs and increased customer attrition.
The transformation of life insurance underwriting will not arrive as a dramatic disruption. It will happen gradually through the replacement of slow, manual processes with intelligent automation.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
