Beyond AI pilots: How insurers are putting AI agents to work

Insurers are using artificial intelligence to speed up underwriting, organize data and handle routine work.

But putting AI into daily operations requires more than adopting new tools. Companies must decide which tasks to automate, how to measure results and where employees need to remain in control.

Those were key points from an AM Best webinar held last week during ITC Vegas 2026.

The panel featured Arun Balakrishnan, CEO of technology provider Xceedance; Adrian Copland, chief operating officer at specialty insurer Ledgebrook; and Manish Chawla, executive vice president and chief information officer at IAT Insurance Group.

The executives discussed how insurers can use AI to improve efficiency while managing the risks of automation, preparing employees for changes in their work and maintaining human oversight.

Insurers look for measurable gains

Balakrishnan said some Xceedance client deployments have produced efficiency gains of 35% to 40%, with some reaching 50%. Companies can use those gains to reduce costs or give employees more time to take on other work, he said.

Balakrishnan described one property underwriting client that used time saved through AI to review more satellite imagery around insured properties.

The additional work helped employees look for potential hazards, including overhanging trees and bushes.

The example shows how an insurer can use AI to expand the work its employees do, rather than simply reduce the time spent on existing tasks.

Chawla said companies should measure more than hours saved. They also need to know whether employees are using the technology in their daily work and becoming comfortable with it.

Balakrishnan cautioned against launching AI projects simply to show that a company is keeping up with the technology. Insurers need clear business goals and ways to measure whether the tools are delivering results.

Underwriters still make the call

One area where the panelists see a clear role for AI is preparing information for employees who make decisions.

Copland said Ledgebrook uses AI to organize data and give underwriters a clearer view of each risk. That can help them assess submissions faster and respond to brokers sooner.

“We want two hands on the wheel at all times for the decision-maker,” Copland said, describing the company's approach to keeping underwriters involved in risk decisions.

Chawla said the right level of automation depends on the type of insurance. Some personal lines tasks may be suitable for automated decisions, he said, but he does not believe AI is ready to take over commercial specialty underwriting.

AI can help move information between teams and reduce delays, Chawla said. But complex commercial risks still require experienced professionals to assess the information and make decisions.

Balakrishnan said insurers should also be careful about how they deploy AI agents, which can perform tasks such as extracting information, checking data, summarizing documents and routing work.

Rather than giving an AI agent broad responsibility for an entire process, companies should set clear limits on each task and decide where human review is needed.

AI raises questions about training and oversight

The push for greater efficiency also raises questions about how insurers will train future employees and retain experienced workers.

Balakrishnan said AI adoption, retirements and difficulties attracting new talent could put pressure on the industry's workforce. He raised concerns about losing the knowledge that experienced underwriters, adjusters and claims professionals have built over their careers.

Chawla warned that cutting entry-level hiring because AI can handle more tasks could leave insurers short of experienced workers later. Entry-level jobs help employees develop the skills needed for more senior roles.

At Ledgebrook, Copland said, recent graduates help colleagues in areas such as finance and claims learn how they use AI. The approach allows employees at different stages of their careers to share knowledge.

Insurers also need to make sure AI tools are used responsibly. Chawla said companies should assess the risks of each application and apply stronger checks when AI influences important decisions. Data quality and potential bias are among the issues insurers need to consider.

The National Association of Insurance Commissioners adopted a model bulletin on insurers' use of AI in December 2023.

It sets expectations for how insurers govern and document their use of the technology and reminds them that AI-supported decisions must comply with applicable insurance laws, including those addressing unfair discrimination.

Balakrishnan also warned that employees should verify AI-generated information before passing it on to brokers or clients. Incorrect information can damage those relationships.

The panelists emphasized that insurers need clear goals for AI adoption, ways to measure its results and safeguards for decisions that affect underwriting and claims.

They also stressed the importance of keeping employees involved and ensuring AI-generated information is checked before it is used.

© Entire contents copyright 2026 by InsuranceNewsNet.com Inc. All rights reserved. No part of this article may be reprinted without the expressed written consent from InsuranceNewsNet.com.

Noah Lopez is editorial assistant for InsuranceNewsNet. Contact him at noah.lopez@innfeedback.com.