Insurers Eye Agentic AI, But Trust in Data Is the Sticking Point

Insurance carriers want AI to run business functions, not just assist them. The barrier is not model capability. It is trust in the data behind the decisions, according to Chris Newman, CEO of ACORD Solutions Group, who spoke at this year's InsurTech Connect conference in Las Vegas.

"I think AI undoubtedly is capable of making very good decisions and evolving things as we're all seeing in businesses today," Newman said. "But the challenge comes with the trust element in terms of can you trust your data."

That question splits into three parts, he said: where the information originated, whether sources are verified, and whether a carrier can account for how a decision was made. Without reliable inputs, a company risks decisions it cannot explain.

Why does explainability matter now?

Regulators are watching. Newman argued that carriers should expect to show evidence and accountability for AI-driven processes, even if regulators stop short of prescribing exact methods.

"The last thing you want from a regulatory point of view is to have decisions being made by a black box and you can't then justify why that's been done," he said.

The compliance angle makes a trust layer a precondition, not an afterthought. "To support the regulatory side and the compliance side, it's really important to have that trust layer," he said. "And then once you've got that base layer, obviously AI then comes into its own from the agentic point of view."

Human oversight stays in the loop. "There are the guardrails that are needed," Newman said. "I think that's going to be required, and it's a sensible thing to have in place rather than to automate everything without those guardrails in place."

What blocks the technical shift?

Legacy systems, and the money sunk into them. Many U.S. insurers have spent heavily on custom-built or acquired policy administration systems, and they will not replace that infrastructure quickly.

"In some respects, they need to preserve their investment in their systems," Newman said. "They've spent a lot of money. They're not just going to change those systems overnight."

ACORD Solutions Group positions itself as the fix. Its role, Newman said, is helping different systems communicate and converting data into trusted sources and formats that other participants can use. The approach accommodates carriers at varying levels of digital maturity rather than forcing migration to one platform.

"We're creating that bridge between the old world and the new world, which means anyone can get involved and digitalize their organization regardless of their level of digital maturity," he said.

The bridge buys optionality. A carrier can build a digital foundation now and swap platforms later. "And then when you're in a position to change your platform or change your system, if you choose to, you've already got that base layer there of digitalization," Newman said. "So, it's really looking to operationalize the standards, bring in that trust layer, but effectively remove a lot of inefficiency that's in the industry at the moment, particularly for insurance companies."

What comes after task automation?

Newman sees the bigger prize beyond automating single tasks. Forward-thinking carriers are redesigning entire workflows on top of trusted data.

"The more forward-thinking carriers are starting to look at their end-to-end operating model," he said. "So they think, 'Okay. We've got the base layer now, so I can now start to really look at how business could be done if it wasn't constrained with some of these legacy processes.'"

Some insurers have already run tactical experiments on smaller transactions. Newman called the move from "hype to actual real implementations" only the beginning. The next decisions: which processes still make sense, which can be redesigned, and what trusted information each change requires.

"Some organizations are starting to step back and say, 'Okay. We understand our business, but this is how some of the processes have been put together. How would we like them to really work? How does the business flow to us from our agents, from our distribution channel?'" he said. "They're starting to re-engineer their operating models so they end up with this new digital operating framework. And I think that's the real exciting opportunity that you can move to."

The industry's near-term agenda is now set: verify the data, keep the human in the loop, then let agentic AI earn a wider role.