Imagine this: the next customer choosing an insurer may never visit its website, see its advertising or speak to one of its advisers. Instead, an AI agent may interpret the customer’s needs, compare products, assess price and coverage, recommend a shortlist and, within agreed permissions, execute the transaction. The human will still carry the risk and pay the premium. But the decisive “customer” in the buying journey may increasingly be a machine.

That changes the competitive logic of retail insurance. Insurers have spent two decades making themselves easy for people to find, understand and buy from. Now they must also become easy for machines to discover, evaluate, trust and transact with.

Agentic commerce: the insurance funnel is giving way to delegation

This is not simply a new digital channel. It is a transfer of work and influence. Consumers already use generative AI to research products and narrow choices. The next step is delegation: asking an agent not only to advise, but to act. And consumers appear increasingly open to that idea. Accenture’s latest Consumer Pulse research finds that 76% of motor and home insurance shoppers are open to an AI-powered personal shopper, with almost one in three prepared to let an agent choose the single best option.

Insurance is particularly exposed. The category is information-heavy, comparison is difficult and many decisions occur at predictable moments such as renewal. Few customers enjoy comparing deductibles, exclusions and contractual language. They have always relied on intermediaries to reduce that complexity. AI changes the economics and reach of that role.

The shift will not happen in one leap. First, agents will explain needs and prepare options. Then they will gather data, request quotes and complete bounded tasks. Eventually, some customers will let an agent recommend and transact within limits they set. High-consequence products will stay hybrid, combining machine preparation with human judgement. But in motor, home and other recurring retail lines, the move from assistance to execution could be faster.

Why now?

Three forces are converging.

• • First, consumer behavior is changing. Natural-language interaction makes it easier to express an outcome than to navigate an insurer’s funnel. In a single prompt, a customer can describe a new home, valuable electronics, a budget and a preference for flexible cancellation. The agent turns that into requirements before any insurer sees the customer.
  • Second, technology is moving from answers to actions. New protocols and payment capabilities are beginning to connect models, tools, product information and transactions. Today’s experience is still incomplete: agents often lack real-time pricing, underwriting inputs and reliable execution. But these are addressable infrastructure gaps, not permanent limits.
  • Third, regulation is becoming a design brief. Consent, authentication, explainability, auditability and accountability will determine how far autonomy can go. Insurers should treat them not only as constraints but as the foundation of trusted delegation, because trust will be the scarce asset in agent-mediated financial decisions.

The consumer gains power, and loses visibility

For consumers, the promise is compelling: less effort, better comparison, more personalized advice and faster execution. An agent can remember preferences, monitor renewal dates, identify protection gaps and challenge inertia.

But delegation creates new dependencies. Customers will need to know whose interests the agent serves, how recommendations are ranked and when commercial incentives shape the outcome. The winning experience will not be the agent that does the most. It will be the one that earns permission to do more.

For insurers, distribution becomes a battle for relevance

The most immediate risk is not that AI replaces every adviser or channel. It is that insurers lose the moment in which customer intent is interpreted and the shortlist is formed.

If an external assistant becomes the place where customers ask, “Am I covered?”, “Am I overpaying?” or “What should I change?”, the insurer risks becoming a product and pricing endpoint behind someone else’s experience. It may still underwrite the risk while losing influence over demand, comparison criteria and loyalty.

Renewals are especially vulnerable. An agent does not suffer from fatigue, forgetfulness or switching friction, and it can re-shop a policy every year. Nearly half of consumers surveyed, 46%, would accept an AI agent switching them to another insurer if it demonstrated a better fit for their needs and budget. Insurers will need demonstrable reasons to stay: superior outcomes, distinctive coverage, or trusted service.

Agentic commerce also creates opportunities: easier access to complex protection, better-prepared human advisers and new demand inside new interfaces. Incumbents may have an important advantage here: 41% of insurance consumers rank a financial institution among their preferred hosts for a consumer-side AI agent, compared with 25% for an AI-native platform, with service and trust among the factors influencing that choice. The strategic choice is not whether to participate, but what role to play: trusted protection orchestrator, differentiated manufacturer, enabling partner, or a deliberate combination by segment and market.

Five priorities for insurers

1. 1. Make the brand legible to machines. Baseline how the company and its products appear in AI-generated answers, and improve the authority, consistency and freshness of the evidence agents use.
   2. Create a product truth layer. Turn coverage, eligibility, exclusions, pricing conditions and contract terms into structured, governed, machine-readable information.
   3. Choose the role in distribution. Decide where to own the interface, empower advisers, partner with platforms or expose capabilities to external agents, by product, segment and market rather than technology fashion.
   4. Build trust into the transaction. Design permissions, human approval, identity, consent, logging and redress from the start, so customers can see what the agent did, why, and how to challenge it.
   5. Put one executive owner in charge. Agentic commerce crosses distribution, product, underwriting, technology, risk and compliance. Shared interest without clear authority will produce pilots, not transformation.

The leadership decision

Insurers do not need to bet the company on full autonomy tomorrow. They do need to avoid being late to the capabilities every plausible future requires: visibility, machine-readable product truth, trusted execution and clear accountability. A practical start is one high-friction journey, such as renewal, quotation preparation or claims intake, run as a governed pilot.

In the internet era, insurers competed to be easy for people to find and buy from. In the agentic era, they must also compete to be selected by machines. Those that move early will not merely sell through a new channel; they will influence how protection is understood, compared and delivered. Those that wait may find the critical customer relationship has already moved elsewhere.