LLMs Are Becoming Insurance Distribution Channels
Insurers Are Now Quoting Inside ChatGPT
A growing number of people start their shopping by asking an AI assistant. Insurers have noticed, and some are now putting their quote journeys directly inside these assistants.
Aviva was one of the first. In April 2026 it launched an app inside ChatGPT that gives a home insurance quote after a few questions, then sends the customer to Aviva's website to buy. In June it added life insurance quotes, according to FinTech Global. Aviva says it will study how customers use the app before extending it to other products, per Aviva's announcement.
Liberty Mutual followed in May 2026 with car insurance quotes inside ChatGPT, starting in seven US states and aiming for more than 40 by the end of the year. Customers answer a short set of questions in the chat, then finish choosing cover and buy on Liberty Mutual's website, according to Liberty Mutual's announcement.
These aren't one-offs. In its Q2 2026 tracker, Evident counted nine new sales and distribution AI use cases among the 30 large insurers it follows. Aviva, Liberty Mutual and Allianz now generate live quotes inside ChatGPT, and AXA is tracking how often LLMs recommend its brand.
%20(1).png)
What Actually Makes It Work
Both launches follow the same pattern. The conversation happens inside ChatGPT. The price comes from the insurer's own systems. The purchase happens on the insurer's website.
The middle step is the important one. Liberty Mutual's app uses its own pricing engine to produce a personalised quote, rather than an estimated price range, according to Reinsurance News. As an eMarketer analysis puts it, insurers now compete on how cleanly their quoting and underwriting systems plug into AI platforms.
To offer a quote inside an LLM, an insurer needs four things in place:
A pricing engine that responds in seconds
The chat can collect answers, but the price has to come from the insurer's real rates, calculated instantly and reachable by another system.
Underwriting rules that can decide on their own
If every quote needs a person to review it, the conversation stalls. Simple cases need to be accepted, declined or priced automatically, with only the exceptions going to an underwriter.
Product answers that come from the insurer's own documents
Customers will ask about cover, exclusions and waiting periods. Those answers have to come from the actual policy wording, not from what the LLM happens to know.
A clean handoff and a record of what happened
The customer moves from the chat to the insurer's website to buy. The quote, the answers given and the source of the lead all need to carry across, for both conversion and compliance. The move to LLMs has already raised questions about compliance and the line between information and advice, as Insurance Business notes.
One thing the integration doesn't solve is discovery. Whether ChatGPT suggests a particular insurer when a customer asks for a recommendation is still out of the insurer's hands, per the same eMarketer analysis. That's why AXA is tracking its visibility in LLM answers.
What This Means for Insurers
An LLM channel sits on top of the systems an insurer already has. If quoting, underwriting and product information are slow, manual or scattered, putting a chat window in front of them won't fix that. If they're fast, rules-based and accessible through an API, adding a new channel like ChatGPT becomes far simpler.
That's the part the Mozart Suite covers. Quote Pro calculates premiums from configurable, version-controlled rates, including channel-specific rates, with maker-checker approval on discounts and a full history of every quote. Underwrite Pro applies configured rules to accept, reject or load a case automatically, and sends anything unusual to a human underwriter. AgentFX answers product questions only from the insurer's own documents, with page-level citations, and recommends plans from a customer profile. It connects to outside systems through an API and works with more than 100 LLM providers. Distribution Pro sets up partner-specific journeys, so a new channel can be configured rather than built from scratch.
For Indian insurers, there's one more consideration. IRDAI's September 2026 consultation paper proposes that prices be shown without asking for personal details, that manipulative website designs be banned, and that every policy be tagged to its seller, according to CareEdge Ratings' analysis of the paper. If those rules are adopted, an LLM channel would likely need to meet them too. We break down the full proposal in our IRDAI explainer.
If you're exploring how your quoting and underwriting setup would support a new digital channel, get in touch.
.svg){kind=small}
.png)