AI Agents: Client Savings, Pressure on Insurers

Personal AI agents are making their way into personal finance, recovering forgotten sums, negotiating bills, or tracking travel credits. Analysts believe their impact could be potentially disruptive for sectors that have benefited from consumer inertia, starting with insurance. Recent stock movements and reimbursements highlight the stakes, but quantifying the impact remains impossible.

Reimbursements, Market, and Accumulating Scenarios

In September, Amazon paid over $845 million in refunds as part of a settlement related to allegations of involuntary Prime sign-ups and cancellation difficulties. In the two weeks following the launch of Muse, shares of Airbnb, Booking.com, and Planet Fitness fell by 12%, 10%, and 17%, respectively. In the insurance sector, one scenario suggests that with 5% of policyholders using an agent and a 10% price drop, $2.8 billion in annual premiums would be affected, but this is a hypothesis rather than a forecast. Rhys Williams, chief strategist at Wayve Capital Management, believes that a 5% to 10% adoption rate would not be enough to change the economics of businesses, but with 40% to 50% of users employing agents, the effects would become notable.

Areas Where Comparison is Opaque Appear Most Exposed

Insurance companies, airlines, banks, internet service providers, and subscription services could be impacted, having previously benefited from customers who do not contest fees, do not compare renewal prices, and forgo complex cancellations. The vulnerability would not be uniform: Craig Moffett, an analyst at MoffettNathanson, believes these agents will have a greater impact where prices are difficult to compare or providers hard to identify. He cites real estate, insurance, and financial services as particularly affected sectors, with insurance being presented as the clearest test. Auto and home insurance premiums in the United States are estimated to be around $560 billion by 2025, representing significant potential if consumers widely use these agents.

What Agents Are Already Doing for Early Users

AJ Stuyvenberg, an engineer at Datadog, used ChatGPT to negotiate with Verizon after receiving a competing offer of $50 per month; after just over an hour of exchanges, his bill was reduced from $94.99 to $69.99, along with a loyalty credit of $47.50, which he estimates would result in an annual savings of $347.50. David Pawlan, growth lead at Merit Systems, used Instinct to monitor the fare of his United flights, then employed AgentCash's StablePhone tool via Claude Code to request credits following a price drop; he reports receiving $73 in refunds on three flights, each taking about five minutes, and believes future personal assistants will be agents capable of proactively handling tedious tasks. Regarding the services offered, Noah Shinn, CEO of Instinct, specifically mentions transaction analysis, subscription detection and cancellation, noting that the agent returns the saved amounts. Meta has also promoted Muse with similar promises. Reported uses range from reducing internet bills to finding forgotten refunds.

Companies on the Defensive, Even Blocking Agents

Observers believe companies will not allow these negotiations to take place without a response. Ben Winters, director of AI and privacy at the Consumer Federation of America, anticipates more complex pathways for consumers and their agents, citing Amazon's decision to block Meta's Muse from making purchases on its site. The standoff could escalate to AI-versus-AI confrontations to retain customers. Tony Soloman, director of insurance intelligence at J.D. Power, expects insurers to also leverage AI to target customers deemed risky and try to retain the most profitable contracts. He reminds us that today, purchasing or modifying a policy still involves human contact, and a shift towards AI would cause insurers to lose control over the customer journey.

Uncertain Impact but Disruptive Potential Mentioned by Several Analysts

So far, no insurer or expert interviewed has provided a reliable estimate of the overall impact. An insurance analyst mentions a certainly disruptive effect, a sentiment echoed by Tony Soloman for his sector. The widespread adoption of AI agents could exert pressure on models relying on customer inertia, as these tools can compare prices and negotiate. They are described as helping consumers recover money while threatening companies that profit from customer inaction.