Personal AI agents go after the revenue companies earn from customer inertia

Business Insider calls its $2.8bn figure "a scenario, not a forecast" ^[7]. No insurer or consumer expert the publication spoke to could put a reliable dollar figure on the impact ^[6]. The scenario starts from roughly $560bn in US auto and homeowners premiums in 2025. It assumes 5% of policyholders use an agent to get a price 10% lower ^[7]. That comes to about 0.5% of the premium pool ^[18]. Rhys Williams of Wayve Capital Management expects the effects to be noticeable at 40% to 50% adoption. At 40%, the same 10% saving would put about $22.4bn at stake, or 4% of premiums ^[19].

The analysts quoted point to insurance. Craig Moffett, a partner and senior analyst at MoffettNathanson, said agents will probably have the biggest effect in markets where buyers struggle to compare prices or to work out who the suppliers are ^[9]. "Real estate, insurance, and financial services all fit the model much better," he told Business Insider ^[10]. Tony Soloman, director of insurance intelligence at J.D. Power, said agents "would certainly be disruptive" for insurers ^[11]. "Right now, to actually purchase a policy or make a change, you still need to contact someone, but if AI replaces that, then the carriers lose control over the consumer journey," he said ^[12].

The share moves offered as evidence of investor pricing came from other sectors. The falls were at Airbnb, Booking.com and Planet Fitness ^[1], and Business Insider wrote that Wall Street had "seemingly begun trading on that thesis" ^[2]. The publication did not report a comparable move in insurer stocks ^[1]. In our view, two weeks of prices in three stocks is thin support for the claim. A window that short cannot separate the Muse launch from anything else that moved those shares.

Williams is the skeptic on timing. "I don't think 5% to 10% of customers are going to be enough to meaningfully change" companies' economics, he told Business Insider ^[8]. The early cases answer a different question: how big the saving per customer is when an agent gets through. AJ Stuyvenberg's Verizon bill fell by $25 a month, about 26% ^[20]. "It's nice to use AI to remove some of the sludge from my life," he said ^[14]. David Pawlan, who works in growth at Merit Systems, said he recovered $73 across three United flights at about five minutes each ^[15]. Noah Shinn, Instinct's chief executive, described what the user sees: "It just sends back the number of what it saved you" ^[4]. Business Insider did not report how many people use Muse or Instinct, so we do not know yet how close any market is to Williams' range.

Companies can answer with friction of their own. Ben Winters, director of AI and privacy at the Consumer Federation of America, said companies could put obstacles in front of consumers and their agents so that they struggle to reach anyone at all ^[17]. That protects renewal pricing while adoption stays low. It also resembles the conduct alleged in Amazon's Prime case: claims that the company enrolled consumers without consent and made cancellation difficult ^[16]. The choice in front of retention teams this quarter is whether an agent on the line gets the same offers a human caller would. If adoption climbs toward Williams' range, we'd expect firms that chose to block agents to face the margin question and the regulatory one at the same time.