I recently sat down with Paul Drake on the Advent of AI podcast to talk about something that's been bothering me for years: the hidden biases in how insurance brokers choose carriers for their clients.

Let me paint you a picture that might sound familiar to anyone who's worked in commercial insurance.
Picture this: It's Thursday afternoon. A broker has just received details about a new bakery looking for coverage. They need to decide which insurers to approach. What happens next?

They might think, "Didn't I place a bakery last month with AXA? That worked out okay, let's try them again." Or perhaps, "If I send three more risks to Aviva this quarter, I'll hit my volume target and earn that higher commission rate." Maybe even, "I'm meeting that underwriter from Zurich for drinks tomorrow anyway, I'll bring this up then."

Sound familiar? These are the hidden biases that influence placement decisions every day across the industry.

The Problem No One Talks About

At Meshed, we're building an AI-native insurance brokerage, and one of the first problems we wanted to solve was this exact issue. Here's why it matters:
We work with over 51 different insurers. Each has multiple products. Their appetites change constantly as markets soften and harden. Some will tell you they don't write restaurants but then happily quote on them anyway. Others claim to love manufacturing risks but decline half the submissions they receive.
Keeping track of all this is practically impossible for a human broker who's already stressed (a recent industry report found 70% of brokers are "either stressed or very stressed" due to time pressures).
I see it firsthand: brokers messaging in team chats asking, "I've got this client that wants to do X, where should I place this?" Even in the Willis Towers Watson Network we're part of, brokers are constantly asking each other where to place certain risks.
The Real-World Impacts
These placement decisions really matter. When a broker doesn't approach the right insurers for a risk:

• Clients might pay more than necessary

• Coverage might not be as comprehensive as it could be

• The process takes longer than it should

And it's not because brokers don't care. It's because they're human, working in a system that makes optimal decisions nearly impossible without help.

How We're Fixing This With AI

At Meshed, we're tackling this problem head-on with our AI approach:
We've built AI agents that continuously gather appetite information from everywhere it exists – emails, PDFs, websites, portal updates – and organize it in a way that's instantly accessible.
When a submission comes in, our brokers don't have to rely on memory or partial information. They simply ask our AI: "I've got a bakery in Manchester with £500K turnover and 5 employees. Where should I place this?"
The AI instantly reviews all 51 carrier appetites and recommends the best options based on comprehensive, up-to-date information – not on who the broker had drinks with last week.
Beyond Just Appetite Guides
What's really exciting is where we're headed next. Soon our system will learn not just from what insurers say they want, but from what they actually accept.
"There might be instances where the appetite guide says 'we don't like restaurants,' but in actual fact, we have been sending them restaurants and they are placing them," I explained to Paul on the podcast.
This reinforcement learning will help us spot trends before they're officially announced, giving our clients an edge in the market.

Why This Matters For Everyone

If you're running a business and buying insurance, wouldn't you want to know your broker is approaching the best possible insurers for your specific needs – not just the ones they remember or have incentives to use?
At Meshed, we believe AI can make insurance broking more fair, more efficient, and more effective for everyone involved. It's not about replacing brokers with AI – it's about giving brokers the tools they need to make truly unbiased decisions.

The technology exists today. We're just putting it to work in an industry that desperately needs it.

Jake Wells is Co-Founder and COO of Meshed, an AI-native insurance brokerage. This post is based on his recent appearance on the Advent of AI podcast with host Paul Drake.