Accenture’s Michael Reilly imagines a world in which AI agents do the shopping for personal lines insurance — a world where price may no longer determine a customer’s decision about which carrier’s coverage to buy.

“The carriers that thrive will not necessarily be the cheapest. They will be the ones that make their real value—claims, service, ease, integrity—as easy for a machine to find and trust as it is for a person to admire,” wrote Reilly, who is managing director for insurance consulting and operations at Accenture, in a report titled, “When the Customer Sends a Bot to Shop.”

In Reilly’s view, that’s going to be happening sooner than most carriers think. In the report (and in the LinkedIn post where CM found it), Reilly referenced an Accenture “Consumer Pulse” survey of 25,590 consumers across 16 countries. The Pulse survey found that almost three-quarters of consumers “would trust a personal AI agent more than their best friend to make a purchase on their behalf.”

That statistic refers to a broad range of purchases, not insurance in particular.

The survey report, later, specifically refers to the purchase of “recurring services,” such as insurance, utilities and telecommunications. For those services, only 12.5% would be open to AI agents “autonomously purchasing” on their behalf. But almost one-third (32%) said they would “ask an agent to make a purchase decision on their behalf within defined boundaries, such as budget and preferences,” according to the survey report, “Talk to My AI Agent: The New Rules of Brand Value.” In addition, 88% said they are open to “collaboration with an AI agent”—working together with an AI to find the best option for recurring purchase decisions like insurance.

Referencing some of the percentages, Reilly highlighted the fact that while personal insurance leaders may think AI agent shopping is a “retail and consumer-goods story,” recurring services (including insurance) rank No. 1 at every level of delegation with an AI agent—ranging from collaboration to handing off the task entirely.

Today, personal lines “is sticky: renewals are automatic, switching is a hassle, and most customers don’t shop unless something forces them to. That friction has quietly protected a lot of books of business,” he wrote, suggesting that change is on the near-term horizon.

While insurance customers today typically weigh “price against a vague sense of brand and a few reviews,” tireless AI agents can compare insurance policies line-by-line differentiating between carriers by price, coverage, claims reputation, service ratings, and ease of doing business, Reilly notes. He adds, however, that even the AI agent can only evaluate “the attributes it can actually find and verify.”

According to Accenture research about all purchases, more than one-quarter of active gen AI users (26%) “have already bought a more expensive item because AI increased their confidence in the decision.”

AI agents “will build on this,” the Pulse report says.

In the world where AI agents shop, carriers who make their value “machine-readable”—exposing verifiable data on claims experience and complaint records, for example—will gain recognition over those relying on slogans about having great claim service, Reilly noted.

“Independent ratings, regulator published complaint data, and transparent cycle-time metrics let an [AI] agent justify a higher premium. Quote-to-bind time, self-service capability, and digital first notice of loss are all things an [AI] agent experiences directly and scores,” Reilly wrote, offering carriers advice to move forward in the world where AI agents shop for insurance policies.

Separately, in late August, JD Power published a U.S. AI Insurance Experience Study, finding that for auto and home insurance specifically, less than one-third of customers (29%) are using AI tools today to research products and coverage, service their accounts, understand coverage before submitting a claim, or shop for a quote or new policy. A media statement about the JD Power study is not specific about the type of tool—gen AI or an AI agent, for example—but highlights the flip side of that statistic for each of the customer activities.

• 71% of customers are still not using AI to research products and coverage.
• 75% are not using it to request quotes.
• 87% are not using it during the claims process.
• 70% are not using it to service their accounts.

“Across all functions, the most common reasons given for choosing not to use AI are lack of familiarity, habit and lack of trust in the results,” says a statement about the study based on 8,352 customer evaluations across 24 insurance brands and eight third-party AI tools fielded in June and July.

Importantly, however, customers who are using AI are starting to pay attention to it. JD Power reported, for example, that 37% of customers who used AI to research products or coverage options changed their policy based on the information they received. When customers used AI specifically to shop for a new policy, 42% purchased a policy as a result.

Before the AI survey was released, Craig Martin, executive director, global insurance intelligence at JD Power, noted that customers now have the potential to beef up their knowledge about how their insurance works with the help of AI. “They can go to AI, which in many ways is a non-judgmental source … I’m not going to look dumb to Chat,” he said during an interview about levels of trust in P/C insurers.

Stephen Crewdson, managing director of insurance business intelligence at JD Power, noted that the U.S. customers participating in the firm’s auto and home insurance shopping studies and reporting that they did use AI in their shopping journeys—again, less than one-third (32%)—said they used it to ask questions about policies and coverages or to compare plans and policies. “Then really to top it off, what we found was those that used AI when they were shopping were more than 1.3-times as likely to switch insurers than those who didn’t use AI.”

“The people using it are asking questions that we know for 26 years, insurers haven’t been great about answering. The needle hasn’t really moved… And those people are more likely to switch by a pretty wide margin,” he said.

Related article: Help Me Understand: How Insurers Build Trust

“We feel like that’s helping them feel more confident: Now I understand insurance better, so I’m more confident to make a decision to switch because AI gave me the information I wasn’t getting in the past from insurers.”

The more recent AI survey found that insurance customers are most comfortable using AI to manage their billing and calculate premiums. They are actually least comfortable using AI to make changes to their current policies or to process claims or manage service inquiries.

What else about AI makes insurance customers uncomfortable?

Thinking about the processes in reverse, roughly half of insurance customers are not comfortable with insurers using AI to handle their claims. “That’s the area where they’re least comfortable, Crewdson said, reporting some of the findings of a separate mid-August 2025 market pulse survey of around 2,000 customers. (“Insurance Customers Keep an Open Mind on Artificial Intelligence, With Some Notable Exceptions.)”

According to that survey, less than one-third of customers are “very comfortable” or “somewhat comfortable” with insurers using AI for “processing insurance claims (reviewing photos, estimating damage, etc.)” or “approving insurance claims.” On the flipside, 47% are “somewhat or very uncomfortable with AI being used to process their claims, indicating a clear boundary between convenience and trust,” the report says.

More acceptable insurer uses of AI, according to the customers surveyed, are activities such as sending billing reminders, sending automated claim status updates and answering basic customer service questions.

“Right now, consumers are not saying don’t ever use it,” Crewdson said, noting that over half are OK with certain functions being influenced by AI. “But they don’t think that they’ll benefit more than the insurer will,” he said, responding to the central question of whether insurers’ use of AI builds or erodes customer trust. Specifically, when asked who will see the most gain from insurance companies integrating AI into their activities, 68% of customers said they believed the insurance company gets most of or all the benefits. Only about one-quarter (26%) said the benefits are shared equally between the customer and the insurer, last year’s JD Power market pulse study reported.

Noting that customers are also uncomfortable when with the idea of insurers using AI to handle is pricing, Crewdson simplified the findings. “They don’t want the money parts to be influenced by AI”—money coming out of the insurer in the form of claims payments or money going from customer to insurer. According to last year’s market pulse study:

• Just 15% of customers believe insurance companies should fully use AI to price their policies.
• One-third (33%) of customers believe AI use in pricing insurance policies should be limited until companies can ensure it doesn’t introduce bias or violate ethical standards.
• Another 30% said AI should be limited to partial use that includes strong safeguards for fairness, explainability and regulatory compliance.

Insurance company representatives who spoke to Carrier Management for our series of articles on trust in the insurance industry described using AI to help human insurance professionals to simplify coverage information and to help explain price changes.

“AI is a tool that our agents, that our reps, our humans use,” said Christine Buxton, VP, head of service transformation at Farmers Insurance. “It’s not replacing what they do because AI doesn’t establish trust right now, right? Our humans do. Relationships do,” she said. Offering an example of one AI-powered tool in use at Farmers—a premium change comparison tool available to agents and customer service reps—she said that it collects needed information about why a customer’s price may have changed at renew, explaining it in simple terms.

“It’s cutting through the complexity that we have in our systems and all of our different data and then pulling that together and providing it to our agent or our rep so that they can then have better and more transparent conversations with the customer,” she said.

“That’s an important thing to call out: Today, AI is making our employees, our agents better so that they are more equipped to have good conversations with the customer to establish trust…,” Buxton said. “We still are in a place where we have like a human at the helm. Even if they have AI to support them or make them better at their job, that human is the person who is establishing the trust and building relationships.”

“We’re just not in a place right now where we expect AI to be able to establish relationships or build trust with customers. We really continue to be a human-focused organization,” she added.

At Nationwide, Sarah Griffin, senior vice president, personal lines product and underwriting, similarly spoke about opportunities to help insurers and agents help customers understand what their policies cover and what they don’t with AI tools.

“To the extent that it can help condense information and simplify that information, as long as we’ve got a human in the loop, and as long as we’ve got the right controls over how we’re using it, we think it’s something that can help clarify information.” Or it can help agents “make sense of all this information, especially independent agents who are getting information from multiple different carriers,” she said.

Griffin spoke to Carrier Management about Nationwide’s mission to build customer trust by sharing “predict, prevent and prepare” approaches with them—including sensor technology—at a time when traditional “repair and replace” measures don’t always meet customer expectations. As for the carrier’s use of AI tools, she said, “We’re still figuring out all of the detailed examples. [But] the way that we look at it and talk about it consistently is as a catalyst for transformation.”

“We want to keep the human in the loop. So, we talk about being people-connected and machine-enabled and AI really being that assistant.” AI is “not the core advisor or the replacement for the judgment,” but AI tools can support that, she said, reiterating the potential for agents to use AI to pull together information that can help them explain coverages to customers.

At JD Power, Crewdson said the levels of trust that customers already have in their insurers factor into their acceptance of those same insurers using AI tools. “Those who are highly trusting of their insurers are about three-times as likely to say they are comfortable with their insurer using AI” for various functions than those who have low trust in their insurers, he reported.