For decades, the Insurance Market has run on a simple, unspoken agreement: you give us the risk data, we give you a premium. How we got from point A to point B was often a mystery – a “black box” calculation buried in actuarial tables and underwriting guides.
But in the age of AI, “trust us” isn’t enough. Brokers need to explain why a rate has changed. Clients need to know how to lower it. And insurers need to prove that their pricing isn’t just a guess… it’s a measurement.
At Cowbell, we are building a “Glass Box” and it starts by rethinking what a risk score actually means.
The Myth of “100%”
When brokers see a Cowbell Factor™ score, the first instinct is often to view it like a school exam: “My client got a 67. That’s a D-plus, right? Why are they failing?”
This is the wrong way to look at modern risk.
I recently had a chat with a new member of our team this week, and explained that a risk score isn’t a static grade out of 100. It is a relative measure against tens of thousands of peers. A score of 67 in the manufacturing sector isn’t a “fail” – it might actually be a solid ‘pass’ relative to the specific digital threats facing manufacturers today.
But context changes everything. If that same business were reclassified as “agriculture” – a sector with different threat vectors and digital footprints – then that same risk profile might suddenly score a 70+.
Why does this matter? Because it moves the conversation from a perceived “bad client” to honing a “specific risk.” It allows the broker to say: “You aren’t being penalised or doing anything wrong. You are being measured against your actual peers, and here is exactly how you compare.”
Confidence in the Code
A common opinion within our industry is that AI is just a fancier black box – a machine making decisions no one can explain. By tackling this head-on (publishing confidence metrics alongside our scores), we don’t just tell you what the risk is; we want to tell you how sure we are about it.
This level of transparency is rare, but it is necessary. If we want brokers to trust output, then we have to show our working. We have to prove that the algorithm isn’t just guessing, but analysing live data from inside and outside the network to build a picture of risk that is far more accurate than a static proposal form.
The “A+” Trap
There is a dangerous misconception though, that if your client has a high security score, then it must mean they’re “safe.”
I often remind people that modern cyber attacks are rarely targeted masterplans by state actors. They are random scans – often by an individual in a bedroom, in a remote part of Europe – looking for any open door.
Companies with fewer vulnerabilities are statistically less likely to be attacked, yes. But having an “A+” security rating doesn’t make you bulletproof. It ignores the human factor. You can have the best firewall in the world, but if an employee clicks a phishing link, the wall comes down.
That is why the “Glass Box” matters. It doesn’t just measure technical defences; it highlights the gaps where human error can creep in. It shifts the insurance product from a “repair bill” to a “roadmap” for better security.
More Handshakes, Not Fewer
Finally, let’s address the elephant in the room: the fear that AI replaces the broker.
I believe the opposite is true. The industry – especially in the London Market – is built on relationships; on handshakes. AI doesn’t replace the handshake, it removes the paperwork that gets in the way of it.
If our AI can handle the “heavy lifting” – the data crunching, the bug spotting, the routine checks – it frees up the underwriter and the broker to do what they do best: solve complex problems.
Think of it like a software engineer using tools like Claude. The AI helps to spot bugs and suggest fixes, but the engineer builds the architecture and ultimately has the last say. The AI makes them better at their job, not obsolete.
In the future, AI won’t mean fewer meetings. It will mean you have the time to have more of them – armed with better data, clearer answers, and a “Glass Box” that builds trust from the very first quote.
