In February 2026, OpenAI approved the first insurer-built application in ChatGPT's app directory. Tuio, a Spanish digital insurer, partnered with AI distribution infrastructure provider WaniWani to embed its entire home insurance quoting process inside a conversation. No portal. No form. No separate tab. A user types a question, the AI collects information through natural dialogue, and a real-time quote from a regulated carrier appears without the user leaving ChatGPT.
Within days, Insurify, one of America's largest insurance comparison platforms, launched a car insurance app in the same directory. Experian followed with its Insurance Marketplace. Last week Aviva, one of the UK's largest insurers, went live with a home insurance quoting app on ChatGPT.
These are live distribution channels, not pilots or proofs of concept.
What just happened and why it matters
OpenAI said in February 2026 that ChatGPT has 900 million weekly active users. WaniWani's own data found that AI was already driving close to 20 per cent of new business for leading digital insurers, and that was before native apps existed. ChatGPT alone was accounting for around 15 per cent of website traffic among surveyed users.
Those users were getting generic answers drawn from static content. Now they can get a real quote.
Raphael Vullierme, co-founder of WaniWani, put it this way: "For the first time, AI can access real offers, quote on behalf of the buyer, and compare coverage in real time. Every insurer will be impacted, whether they've built an AI app or not."
That last sentence is the one worth sitting with. The distribution shift does not require you to have built anything. It is already happening to you.
The buyer is changing before the sale begins
The traditional insurance buying process had a clear shape. The customer experienced a triggering event: a new car, a house purchase, a business launch. They searched, compared, called, or asked a broker. The insurer or broker controlled the first meaningful interaction.
That shape is changing upstream.
When a customer opens ChatGPT and asks "what insurance does a small business owner in Sydney actually need?", the AI answers. It frames the categories of risk, describes the coverage types and, if an insurer has built an app, begins quoting. By the time a human intermediary enters the picture, value has already been framed, expectations have been set, and in some cases a purchase has been completed.
Digital Insurance, the US trade publication, described the structural consequence in March 2026: "AI framing can outweigh marketing investment, SEO placement, or channel preference." The first explanation of value is now being delivered by an AI interface, not by the insurer, its agents, or its distribution partners.
This has a specific implication for insurers who have invested heavily in brand differentiation. Fine-grained product distinctions (the specific policy conditions that separate a good product from a mediocre one) get compressed into broad labels inside an AI conversation. "Good value." "Expensive." "Confusing." The AI decides which label applies, based on how your product is structured and how accessible that structure is to a machine.
The real enabler is not the AI. It is what sits behind it.
The thing that made Tuio and Insurify's ChatGPT apps possible was not artificial intelligence. It was their API infrastructure.
Tuio could distribute inside ChatGPT because its entire rating and quoting engine was already exposed through real-time APIs. The AI interface is the last mile. The infrastructure is what made it reachable.
This is a pattern across every significant AI distribution development in insurance right now. McKinsey's February 2026 analysis of AI in insurance concluded that the industry is most likely to be reshaped rather than disintermediated, but that reshaping depends entirely on the readiness of the systems behind the distribution channel. BCG, writing in January 2026 on P&C insurers and AI, noted that AI-assisted quote intake and pre-fill can cut time-to-quote by 30 to 40 per cent, but only for insurers whose systems can respond in real time via API.
The 2024 Insurance API Index found that 86 carriers and MGAs were already offering API-enabled products at that point, and many more insurers have since embedded APIs into their digital operations. That sounds encouraging. But embedded APIs for internal workflows are not the same as open, real-time APIs capable of delivering a rated quote to a third-party distribution platform in under a second. Most insurers have the former. Very few have the latter.
What this means for Australian and New Zealand insurers
The Tuio and Aviva apps are live in Europe. Insurify and Experian are live in the United States. Australia is not there yet, but the infrastructure precedent has been set, and it will arrive.
The distribution dynamics here are somewhat different. Australia has a stronger intermediary culture, particularly in commercial lines and small business insurance, and the regulatory environment around personal advice creates friction for fully automated purchasing in some product classes. But these are constraints on speed, not direction. The underlying dynamic is AI sitting between the insurer and the customer, framing value before a human is involved. That is a structural shift in how information flows, and it won't stay in North America and Europe.
For Australian insurers and coverholders, the more immediate question is not "should we build a ChatGPT app?" It is "could we, if we needed to?" If your rating engine is locked inside a legacy platform, your quote flow requires human intervention or your product data is not structured in a way that a machine can consume, the answer is no, and you are not ready for a distribution channel that is already operating in other markets.
The businesses that get ahead of this are the ones making infrastructure decisions now that they will not need to execute until next year. Getting your API layer ready keeps your product reachable as each new distribution surface appears: AI agents, embedded insurance partnerships, automated procurement systems. A ChatGPT app is only one of them.
What brokers and MGAs should be doing
The instinct in the intermediary market is often to treat AI distribution as a threat of disintermediation. That instinct is understandable but partially misplaced.
McKinsey's analysis is consistent with the broader consensus: AI will reshape, not eliminate, the broker's role. PwC's Insurance 2030 outlook sketches a future in which quotes typically come from AI interactions, while customers who still want advice from a person, most likely for big-ticket commercial and group coverage, turn to advisers with bespoke approaches. The volume business may move toward AI-mediated channels. The judgment-intensive business will not, at least not soon.
But brokers who assume their relationships will protect them without updating their infrastructure will find themselves in a deteriorating position. A customer who researches inside ChatGPT, receives a competitive quote and completes a purchase without speaking to anyone is a customer no broker has touched. That customer's next renewal starts from a different baseline.
The practical steps are not dramatic:
Map where your products are discoverable. If a customer asks an AI assistant about the type of insurance your business writes, does anything useful come back? If not, that is the first problem to solve. It is largely a content and data structure problem rather than a technology problem.
Understand your API readiness. Can your systems deliver a real-time quote to a third-party platform? Embedded insurance, AI distribution, partner integrations, and comparison platforms all depend on the same capability. If you do not have it, build it or partner with someone who does.
Own your specialism. AI apps are starting with commoditised personal lines: simple, high-volume, low-advice products, chosen because they are the easiest to structure for machine consumption. Niche commercial lines, specialty risks, and products requiring genuine underwriting judgment are materially harder to automate. That is where experienced MGAs and specialist coverholders have a durable advantage, but only if they are actively cultivating it.
The question to ask your leadership team this week
Here it is, plainly: if a customer asked an AI assistant for the type of insurance your business sells, and that AI had the ability to quote and bind in real time, would your product be available?
If yes, you are ahead of most of the market.
If no, that gap is closing faster than most people in this industry realise.
The interface is not the product. It is now the first thing the customer sees, though, and in a growing number of cases the only thing they interact with before a policy is purchased. Building for that is the next five years of distribution strategy.
Sources: Reinsurance News / FinTech Global (Feb 2026), Insurify press release (Feb 2026), Experian press release (Feb 2026), ResultSense / Aviva (Mar 2026), Digital Insurance (Mar 2026), McKinsey Insurance AI Analysis (Feb 2026), BCG AI-First P&C Insurer (Jan 2026), Insurance API Index (2024), PwC Insurance 2030 (2023), TechCrunch (Feb 2026), WaniWani distribution data (2026)