Two years ago, we invented a man called Ian Trelom.

Ian was the founder of an independent games studio called InaudibleMirage. His studio was developing a title called Insurance Simulator 2024, a management game where players built and ran their own insurance company.

We made screenshots, logos, feature lists and a launch website. Ian even supplied a quote about how excited he was to be working with Genasys who had consulted on and sponsored the game.

There was only one problem.

Ian Trelom did not exist.

Ian Trelom is an anagram of "I'm not real", a detail we were certain somebody would spot within the hour. Nobody did. Not on launch day, not when we revealed the joke, and not once in the two years since.

The whole thing was an April Fool's prank.

At least, that was the original plan.

The nagging thoughtThe joke kept describing a real game

The further we went building the joke, the harder it became to ignore something. Every feature we invented for the fake game sounded like something we would actually want to play. The more detail we added, the clearer it got: running an insurer has most of the ingredients that make a management game work.

Most people outside the industry picture insurance as paperwork, forms and regulation. Those of us inside it know it is really a long exercise in decision-making under uncertainty. Every day is a series of judgements made on incomplete information. Which broker do you trust? Which risks do you write? Which warning signs do you investigate, and which do you wave through because you are busy? The consequences rarely show up until months or years later, by which point you have made a hundred more.

And that's what makes insurance so utterly gamifiable.

So the joke turned into a question. Could you build a real simulation around the actual job? Could you capture the tension of balancing growth, profit, compliance and reputation, and make something an underwriter would recognise while a graduate could still pick it up cold?

In 2024, those questions were academic. Building the game would have meant a development team, artists, designers and a budget nobody was going to sign off for what began as a marketing gag. So we let it go.

Then the tools caught upThe joke stopped needing to be a joke

Over the two years since Ian Trelom first appeared, AI-assisted development demolished the barrier we had run into. Tools that barely worked in 2024 made things possible for one person that used to need a small studio. We started prototyping, mostly to see if we could, and realised something faintly dangerous.

The joke no longer had to stay a joke.

So we built it. Insurance Simulator 2026 is now a real game, free to play in your browser, made by our marketing team, which is one person, using the latest generation of AI coding tools alongside Midjourney, ChatGPT and Photoshop. There is still no studio. It turns out we no longer needed one.

The job, as a gameEvery decision looks simple from outside

Remember those questions. Which broker do you trust, which risks do you write, which warning signs do you investigate. The game does not ask them in the abstract. It puts them in front of you and makes you answer.

Which people do you hire?

Your first serious candidate is Rohan Ghosh. Fifty-nine, senior underwriter, ACII, and on paper superb: risk assessment 100, pricing accuracy 100. He wants £88,000 and a 22% bonus, and he is on probation, and buried in his profile is an integrity score the game will only tell you sits somewhere between 16 and 66. You learn which end when it is far too late to matter.

Which brokers do you trust?

Mercer & Ridley want an agency. Thirteen years trading, £2.4m of financial lines, run by a market fixture called Toby Stafford, FCA record showing three complaints, all resolved. The biography is warm right up to the sentence that mentions, in passing, that some underwriters have flagged inconsistency in how they present renewal information. You get three steps to decide whether that one line is a red flag or just how the sausage is made.

Which risks do you write, and at what price?

A clean risk arrives. No claims in three years. The market has come back between £1,300 and £1,600, and the broker lets slip the client wants to pay about twelve hundred. You can price it on instinct, spend £300 on a credit report, or £432 to call the broker and fish for more. Your decision confidence, the game notes helpfully, is "Limited". Welcome to underwriting.

And when the press call?

Insurance Times asks for a word. Katie Scott, thirteen seconds a question, four answers, each a slightly different flavour of overconfident. There is no right answer, only how you will read in print. Play it careful and you get the headline "Market's newest chief plays it close to the chest", which is the polite version of saying you gave them nothing.

Read that write-up properly rather than clicking past it, and the sidebar is running its own headlines.

Nobody was meant to read those, the same way nobody was meant to unscramble Ian Trelom. We keep hiding things in the margins to see who is paying attention. Mostly the answer is nobody, which is somehow both disappointing and entirely the point.

LosingThree ways it ends, all of them your fault

You do not really win Insurance Simulator. You postpone losing it. There are three exits, and anyone who has worked in this market will recognise every one.

Here is what the game gets right, and the reason it stopped being a joke. Every one of those endings traces back to a decision you made somewhere else entirely, a few years earlier, feeling quite pleased with yourself. The cheap hire. The broker whose bio you skimmed. The quote you rushed. Nobody loses an insurer in a single dramatic moment. You lose it slowly, in small sensible-looking steps, which is exactly how it works in the building the rest of us go to on Mondays.

How we built itNinety per cent thinking, ten per cent code

The whole thing was made by our marketing team, which is one person, using the latest generation of AI coding tools alongside Midjourney, ChatGPT and Photoshop. The tools have moved fast enough that the coding is no longer the hard part.

The hard part, and close to ninety per cent of the work, was the thinking. Deciding how the business should behave. Working out which corners of the insurance industry to model and which to leave out. Designing the gamification so that a loss ratio or a reserve means something on screen. Then testing it, modelling it, breaking it and tuning the numbers until losing money feels like losing money. A game where nothing you do has consequences is not a game, it is a form, and getting that balance right took far longer than the code ever did.

What comes nextThe roadmap is a hoax from 2024

Delegated authority. Mergers and acquisitions. Deeper reinsurance mechanics. And the London Market, because no insurance game would be complete without it.

Go back and read the feature list of the game that never existed, and there they are: delegated authority, M&A, the lot. The roadmap for the real game is a hoax press release we wrote two years ago. We are entirely comfortable with that.

Ian Trelom never existed. Two years on, everything he was pretending to sell does. And it turns out running an insurer is every bit as hard as we suspected.