What will the first AI life insurance broker look like?

This article was first published on LinkedIn on 8 September 2026.

New technology is often described in the language of the system it is about to replace. Every time you "cc" a colleague, you are sending them a "carbon copy". The convention comes from early twentieth century office correspondence, when a typist placed carbon paper between two sheets and produced a memo and its copy. The phrase survived because it described a relationship, not a mechanism.

The protection industry is trying to do something similar with its ambitions for AI. Ask the market what an AI broker is for and the answer is usually a digital sales assistant: one that can answer questions, produce quotes, collect information and prepare a consumer for a conversation with a human adviser.

It is a valuable vision, but it still describes AI through the structure of today's sales process, with the ultimate aim of getting a consumer onto the telephone.

The more ambitious possibility is that the natural language interface allows the process to begin somewhere else entirely. The conversation can be led by the consumer and shaped around their circumstances, concerns and intention.

An AI life insurance broker will not be a chatbot bolted onto an existing form. It will be a regulated, persistent distribution system that discovers need, develops intention and completes the sale. AI will reshape the relationship between protection need, protection intention and protection process.

Why the protection market will initially resist this version of AI

For the last twenty years, protection has been organised in one direction. Reinsurers decide which risks they want to accept. Insurers manufacture products they believe will be profitable and build a fixed process for quotation, underwriting and quality management. Distributors are paid to find consumers for those products and persuade them to enter a telephone sales conversation.

This model is famously expensive, but it is also predictable. The parties involved can forecast acquisition costs, cancellations and long term margin with reasonable confidence. But what the current model has struggled to do is materially grow the market.

Even as the protection customer journey has fallen behind other insurance lines, the industry has clung to what it can control and its technology reinforces that direction. Forms, portals and telephone processes expect consumers to answer predetermined questions in a fixed order.

So much money and time has been invested in the current sales process that AI will have to prove, through live deployments, that a different model can produce better consumer and commercial outcomes.

Natural language reverses the direction

Consumers are rapidly adopting natural language conversations with AI. According to Ofcom's Adults' Media Use and Attitudes 2026, "54% of UK adults now use tools such as ChatGPT, Copilot or Gemini, up from 31% in 2024."

Whether someone is researching life insurance or looking for a recipe, conversational AI begins by listening to what they are trying to achieve. It can ask follow up questions, interpret context and decide what information or action should come next before a product has been introduced.

In protection, that means understanding what has changed in a consumer's life, what they are worried about, what cover they already have and whether they genuinely intend to act.

If the answer is "not yet", the conversation does not have to be treated as a failed lead. With the consumer's permission, it can continue when the need becomes clearer or the intention becomes stronger. Only once that intention has formed does the AI broker introduce the appropriate product, process and provider.

AI will do more than make the existing protection journey faster. It will reverse the direction in which the industry works. The journey will begin with the consumer's circumstances and assemble the right product and process around them.

What capabilities will an AI broker have?

It will always be available

A consumer who starts thinking about protection at 11pm will be able to discuss what they are worried about immediately.

The AI broker will answer the question the consumer actually asked, help quantify the potential exposure and explain the available options. If the consumer is not ready, it can remember the interaction and, with permission, resume the conversation at a more appropriate time.

Persistence matters because protection intention rarely forms in a single session. A house move, a new child or a change in employment might create the need, but the decision to act can develop over days, weeks or months.

It will know when a human is needed

AI brokers will not remove people from every protection journey.

Depending on the proposition, an AI broker might provide information, complete a non advised sale or recognise that the consumer's circumstances require human judgement or a personal recommendation.

When an adviser is needed, the handover should not force the consumer to start again. The adviser should receive the conversation record, the relevant context and a clear explanation of why the case was escalated. Advisers will ultimately spend more time on the cases that benefit from their expertise.

It will be built around the Consumer Duty

A distribution system that talks to consumers at scale must be designed around Consumer Duty principles from the outset, rather than audited into compliance afterwards.

AI creates new risks because a poor explanation or decision can be repeated across thousands of conversations. But it also creates an opportunity for stronger and more consistent oversight.

Every interaction can be recorded, structured and assessed against defined conduct rules. Firms can monitor consumer understanding, identify emerging harm and record when and why a conversation was escalated.

The system will still require clear accountability, rigorous testing and human oversight; properly governed, an AI broker could provide firms with much richer evidence about consumer support, understanding and outcomes than periodic reviews of a small sample of telephone calls.

It will fit naturally into bancassurance

Some of the earliest AI brokers are likely to appear within existing financial relationships.

Inside a banking app or mortgage journey, an AI broker can begin a protection conversation at a relevant moment without forcing the consumer into a disconnected lead form. With the appropriate permissions, it can use information the consumer has already provided to reduce repetition and establish context.

The conversation might begin with a mortgage, a change in household income or a question about financial resilience rather than an explicit request to buy life insurance. This all makes bancassurance particularly well suited to conversational protection distribution.

It will require a different technology architecture

Natural language is less orderly than a form. Consumers change subjects, provide incomplete answers, ask for explanations and introduce information before the process expects it.

An AI broker will need a technology architecture capable of interpreting those inputs while still applying deterministic product, underwriting, conduct and escalation rules. It must know when it can respond freely, when it must follow a defined process and when it must stop and involve a human.

Some brokers will build this technology themselves. Others will work with specialist platforms or integrate conversational systems into existing insurer and portal infrastructure. The defining feature will not be ownership of the stack, but the ability to support an unstructured conversation within a controlled and auditable process.

It will produce distribution intelligence

Every protection conversation contains information about demand.

An AI broker will know which questions consumers ask, which concerns prevent them from acting, which features they value and where existing products fail to fit their circumstances. Today, much of that intelligence is lost in telephone recordings, abandoned forms and disconnected CRM notes. AI can structure it as the conversation happens.

With enough evidence, AI brokers will be able to identify underserved segments and show insurers and reinsurers the need, product design and price point consumers are looking for. Distribution will not simply sell products manufactured elsewhere, it will increasingly influence what gets manufactured.

When can we get started?

Now.

The first AI success stories are likely to come from insurance businesses that apply the protection market's culture of acquisition testing and learning to the entire distribution journey.

They will use natural language conversations to understand how consumers express need, when intention forms and what prevents people from acting.

The significance of the first policy sold by an AI broker will not simply be that a machine completed a sale. Far more important will be that after years of building journeys around products and processes, the protection industry will have organised distribution around the customer.

Today, the industry defines the protection process and searches for consumers willing to enter it. Tomorrow, the consumer will begin with their own circumstances, and the protection process will assemble around them.

Alain Desmier is co-founder of Subcontext.