Personal AI agents threaten companies profiting from consumer inertia

Personal AI Agents Help Users Review Subscriptions and Negotiate Service Prices

Personal AI agents are being used by some consumers to review recurring charges, compare service options, and request adjustments to bills or credits. These tools scan transaction records, identify active subscriptions, and in some cases interact with providers on behalf of users.

Noah Shinn, chief executive of Instinct, described one such agent on the podcast “Invest Like the Best” last week. He said the system can examine bank transactions, locate unused subscriptions, and initiate cancellations, returning a summary of the amounts involved. Meta has presented its agent, Muse, with similar capabilities for handling routine financial tasks.

Several industries, including insurance, airlines, banks, broadband providers, and subscription services, have historically retained customers who do not regularly compare renewal rates or complete cancellation steps. Analysts have noted that wider use of personal AI agents could alter retention patterns in these sectors.

Stock prices of certain companies moved lower in the two weeks after Muse became available. Shares of Airbnb declined 12 percent and Booking.com shares fell 10 percent during that period. Planet Fitness shares dropped 17 percent over the same interval.

No insurer or consumer-sector analyst contacted by Business Insider supplied a precise estimate of aggregate financial effects. Tony Soloman, director of insurance intelligence at J.D. Power, stated that personal AI agents “would certainly be disruptive” for insurers. He added that if AI systems handle policy purchases or modifications, carriers would have reduced direct oversight of customer interactions.

Early examples of use have been reported by individual users. AJ Stuyvenberg, an engineer at Datadog, directed ChatGPT to negotiate with Verizon after receiving a Comcast fiber offer of $50 per month. After roughly one hour in a Verizon chat session, the agent obtained a monthly rate of $69.99, reduced from $94.99, along with a $47.50 loyalty credit. Stuyvenberg calculated the changes produced $347.50 in savings over one year.

David Pawlan, who works in growth at Merit Systems, employed Instinct to monitor United Airlines fares and then used AgentCash’s StablePhone tool via Claude Code to request credits when prices declined. He reported recovering $73 across three flights, with each request taking about five minutes.

Craig Moffett, a partner and senior analyst at MoffettNathanson, said the tools are most relevant in markets where prices are difficult to compare or where identifying alternative suppliers requires substantial effort. He listed real estate, insurance, and financial services as sectors that match this description.

Insurance has been cited as one area that could see measurable activity. U.S. auto and homeowners insurance premiums totaled approximately $560 billion in 2025. A hypothetical case in which 5 percent of policyholders used an agent and obtained a 10 percent reduction would place $2.8 billion in annual premiums at risk. Analysts emphasize that this remains an illustrative scenario rather than a projection.

Rhys Williams, chief strategist at Wayve Capital Management, said adoption levels would need to be higher for broad economic effects. “I don’t think 5 percent to 10 percent of customers are going to be enough to meaningfully change companies’ economics,” he said. “If 40 percent to 50 percent of customers are using agents, then the effects will be noticeable.”

Some companies have already faced regulatory or settlement costs related to enrollment and cancellation practices. Amazon has issued more than $845 million in refunds as of September under a settlement concerning allegations that consumers were enrolled in Prime without explicit consent and that cancellation procedures were complex.

Businesses are expected to respond to increased use of consumer-side agents. Ben Winters, director of AI and privacy at the Consumer Federation of America, said providers could introduce additional steps that limit access for both consumers and their agents. He cited Amazon’s decision to restrict Meta’s Muse agent from completing purchases on its platform.

Soloman of J.D. Power noted that insurers may also deploy AI systems to identify customers at risk of switching and to offer retention options proactively. This could lead to automated interactions between consumer agents and company systems.

The article continues with additional reporting on deployment timelines, technical limitations of current agents, and statements from other analysts at firms tracking technology adoption in consumer services. Further details include the number of early-access users reported by several agent developers and the specific data fields these systems are permitted to access under existing terms of service agreements.