From Lloyd's Coffee House to ChatGPT: 300 Years of Insurance Distribution, and why the next shift is the biggest yet.
In 1688, Edward Lloyd opened a coffee house on Tower Street in London. It wasn't remarkable by the standards of the time. Coffee houses were everywhere in Restoration London — places where merchants, sailors, and traders gathered to exchange information and argue about politics.
What made this specific coffee shop different wasn't its oat-milk alternatives or latest spin on the Matcha Latte, but rather its clientele. Ship owners, cargo merchants, and the financiers willing to underwrite the risk of oceanic trade gathered there because Lloyd's was where you went to get intelligence about ships, routes, and the relative safety of particular voyages. The information was good, and the people who needed it knew where to find it.
Out of those conversations, almost incidentally, came the modern insurance industry. Underwriters would literally write their names under a description of the risk they were willing to cover, hence the term.
The coffee house became the exchange, the exchange became Lloyd's of London, and Lloyd's of London became, for three centuries, one of the most important insurance markets in the world.
The origin story matters because it contains something true of insurance distribution ever since: the product goes where the information is, and the consumers often follow suit. Lloyd's didn't necessarily invent insurance, but they created a place where the people who needed it and the people who could provide it could find each other — they nailed the marketplace dynamic and were one of the first to really solve liquidity. That's what every insurance distribution campaign has attempted to do ever since.
The era of the insurance broker
For most of the three centuries that followed, insurance distribution equated to brokers. Human intermediaries, or as we now call it, a 'human-in-the-loop', who understood the products, the risks, and the clients and who made their living connecting the three. For complex commercial risks, the broker's judgment was genuinely irreplaceable and remains valuable in many scenarios. For personal insurance, the model can feel expensive, slow, and inaccessible to those whose risk is too small to be worth a broker's time.
True democratisation of insurance access had to wait for a different kind of infrastructure.
The telephone and the direct revolution
Direct Line launched in 1985 without branches, brokers, or intermediaries. It took calls, gave quotes, and sold policies over the phone. For its time, this was next-level innovation and disruption through new distribution strategies. Prices were lower because the cost base was lower, and as always — the incumbents were dismissive. The broker community was (understandably) equally as hostile. Neither adequately anticipated what Direct Line had actually figured out: most personal insurance customers didn't necessarily need advice; they simply needed a price and a policy — the necessary information to self-compare and make an educated decision based on their specific needs.
By the early 1990s, direct insurance had started becoming the norm. Churchill, Privilege, and a wave of imitators had followed, and a new pattern was set.
The common trend, with any new technology, is that it creates a new distribution surface, early movers build advantages that are difficult to dislodge, and the insurers who wait too long ultimately find themselves trying to catch up within a market they didn't shape.
The internet and the aggregator
Confused.com launched in 2002. GoCompare followed in 2006. Within a decade, the aggregator model had fundamentally restructured personal lines. Price became the primary competitive lever while the strength of the brand became a means of surviving on aggregator platforms.
The aggregator era also produced some of the most creatively ambitious insurance marketing in history. In our recent article, The Insurance CMO Playbook For The AI Era, we highlighted Comparethemarket's Aleksandr Orlov, alongside a handful of other innovative guerrilla, OOH, and distribution-level campaigns. The best insurance campaigns of this era worked because they met consumers exactly where they were, which, at the time, was on a comparison website. And in order to rank highly, that website had to be ahead of the curve with the latest SEO strategies.
The embedded era of insurance
The most recent shift before the current LLM-oriented one was slightly quieter but no less significant. By Miles built pay-by-mile car insurance natively integrated with connected car data. Laka built collective cycle insurance that spread across communities. Urban Jungle built renters insurance simple enough to sell inside a fifteen-minute onboarding flow.
The logic was elegant and aligned with a clear hypothesis: if the consumer was already on a platform booking a flight, buying a car, or signing a lease, the insurance need was right there and would be a logical next step for the consumer. Why redirect them, adding unnecessary friction, to a comparison site? The product and the distribution became, in the embedded model, effectively the same thing.
So what's the insurance distribution shift that's happening right now, in 2026?
As we've mapped out, each distribution era was enabled by a specific technology: the coffee house information network, the telephone, the internet, and the website-based API. Each time, the technology created a new surface where consumer attention and insurance intent could meet. Each time, the insurers who understood what the new surface required built advantages that compounded their distribution, from awareness all the way through to retention.
The technology enabling the current shift is the large language model (LLM). Powerful software like ChatGPT, Claude, or Perplexity handles billions of queries per week. A growing proportion of these queries are increasingly purchase-adjacent, as businesses and consumers opt into the LLM as their go-to assistant for daily queries, deep research, and personal advice.
Questions about what insurance is needed, how much cover is appropriate, and which products exist for specific risks. These are the opening moments of a buying journey, and they're increasingly happening inside AI assistants rather than search engines.
Aviva's recent ChatGPT launch is the clearest signal yet that this new window is open. What makes this shift different from its predecessors isn't the scale of the technology or the speed of adoption — both are incredibly impressive. It's the nature of the interface. Every previous distribution channel was a destination the consumer chose to visit. The AI assistant is something quite different. Online "Chat" has become the golden thread of global communication.
Whether that's between friends and family on WhatsApp, amongst colleagues on Slack, or when conducting research via ChatGPT, we're all — as humans — familiar with, have, and interact with every day. It's one of the most intuitive forms of communication in the modern world, and how most consumers and businesses converse. And now, with innovative, AI-native software like Malcolm, insurance companies can leverage these conversations to meet their prospective customers where they are.
For that conversation to result in an accurate, compliant, bindable quote, the underlying infrastructure has to exist. That infrastructure is being built today, and is already being rolled out to the insurance winners of tomorrow.
If the last 300 years have taught us anything, it's that the window between early mover and laggard is always shorter than it looks.
Malcolm is building the compliant infrastructure layer connecting insurance pricing engines to AI distribution channels. If you're thinking about what this shift means for your distribution strategy, we'd be glad to talk.