Your next insurance customer may never visit your website, scroll through your coverage options or personally fill out a quote form. Their AI agent might do it for them.
That stopped being a hypothetical this month.
Meta launched Muse on Sept. 8 as a personal AI agent that can open a browser, fill out forms, negotiate and take action on a user's behalf.¹ One early user uploaded his existing auto insurance policy and asked Muse to find equivalent coverage for less. According to the user, Muse found a policy that saved him $3,500 a year, purchased it and canceled the old policy in about five minutes.²
And Meta isn't alone. On Sept. 29, OpenAI introduced Dots, always-on AI agents that can use their own cloud computers, connect to thousands of applications and carry out multistep tasks on a user's behalf.³ The launch marks another step toward a world where consumers delegate more of their shopping, research and transactions to AI.
Consider how different that transaction was from the insurance shopping experience we've spent decades building. The consumer didn't visit multiple carrier websites or spend an afternoon filling out forms, and he didn't need to call an agent. His software went shopping for him and completed the transaction.
Every carrier and agency in P&C now needs to think about a new kind of customer: an AI agent acting on behalf of a human.
The internet is getting a new user
Insurance is only one place where this is starting to happen.
Agentic commerce is emerging across the economy. Stripe has integrated its Link wallet with Muse, allowing the agent to transact at more than 1 million businesses that accept Link and use single-use virtual cards elsewhere.4 Visa and Mastercard are also building infrastructure designed for transactions initiated by AI agents.5 6
BCG has specifically identified this shift in insurance, describing AI assistants as a potential new "front door" through which consumers discover, compare and buy coverage.7 Deloitte found similar consumer interest: In the 2025 Insurance Barometer Study it cites, 55% of respondents said they would use an AI tool to shop for life insurance.8
Until now, most digital insurance distribution has assumed a human is sitting behind the screen. That person gets tired, abandons forms, doesn't necessarily check every carrier and may choose a familiar brand even when another option is cheaper.
An AI agent approaches the process differently. Give it a goal, and it can potentially compare quotes, decide which carriers to evaluate, complete forms and continue working until the task is done.
The implications for insurance distribution are enormous.
Amazon just showed us the other side
Companies are already deciding how much access they want to give these agents.
Amazon recently blocked Muse from shopping on Amazon.com. The company said Muse had accessed its store without authorization, did not identify itself while browsing and raised concerns about customer credentials, privacy and security.9
At the same time, Amazon is developing its own agentic shopping capabilities. Alexa for Shopping can research products, compare options and make purchases on a customer's behalf.10
That tension offers an early glimpse of a much larger issue: Companies need to decide which agents can interact with their systems, what those agents are allowed to do and how authorization should work.
Insurers will face their own versions of those questions.
Can an outside AI agent request a quote on someone's behalf? How does the carrier verify the customer's authorization? Can the agent access policy information, make a coverage change or file a claim? What happens when it calls the contact center? Identity, consent, disclosures and audit trails all become more complicated when software is acting for the policyholder.
These questions are arriving quickly, and most insurance technology road maps weren't built with them in mind.
Your contact center may hear from AI, too
The implications extend beyond websites and quote forms. When an online workflow doesn't provide what an AI agent needs, calling the company becomes another way to finish the task.
We could soon see an insurer's AI talking directly to a consumer's AI, with one agent representing the policyholder while another navigates the carrier's systems, applies business rules and completes the transaction.
But consider what happens when thousands of consumer AI agents start calling insurance companies on behalf of their users.
Some may be gathering quotes, comparing coverage or checking prices with little indication that the consumer is ready to buy. Others may be trying to complete routine servicing requests. Without a way to distinguish AI-generated inquiries from human callers, insurers and agencies risk having their licensed agents and customer service representatives spend valuable time interacting with bots instead of customers who need their expertise.
The answer is to let AI handle AI.
Insurers and agencies need technology that can identify when an inbound interaction originates from a consumer AI agent and route it to an AI system equipped to respond. That system can handle routine inquiries and complete authorized transactions, while escalating complex or consequential cases to a human representative.
This approach enables insurers to deliver information faster, elevate the customer experience and accommodate a growing volume of AI-generated interactions without allowing AI traffic to overwhelm their human teams.
The customer is still human
Of course, the policyholder hasn't disappeared. The human still has the need, owns the policy and ultimately lives with the consequences of the decision.
That makes trust, consent and control even more important as software gains the ability to act with real money and make consequential choices. Recent Visa research illustrates how much work remains: While consumers are increasingly using AI during shopping, only 23% of U.S. consumers surveyed said they trust generative AI to handle payment transactions on their behalf.11
For insurers and agencies, the challenge is to accommodate these new digital intermediaries without compromising the experience of the people they represent. That means knowing when AI is calling, understanding what it's authorized to do and maintaining appropriate safeguards and human oversight.
The next wave of insurance customers may arrive through AI agents. Insurers and agencies need to be ready to serve them without pulling their own people away from the conversations where human judgment, empathy and expertise matter most.
