Don’t fool yourself: artificial intelligence (AI) could take your insurance job, a brokerage CEO warned during a recent industry event.
“There were some conversations yesterday in this room about how AI is not taking our jobs, AI is just making our jobs more efficient,” Stephen Billyard, CEO of Billyard Insurance Group (BIG), said during a two-day Reuters Events conference. “I think that’s something we’re just saying to make ourselves feel better about what AI is actually going to do.
“AI is taking our jobs, and it has taken a whole lot of underwriters’ jobs out of our business. So, I think we need to be honest about that,” Billyard said during the Future of Insurance 2024 event in Toronto on Dec. 13. “Certainly, that’s not what we’re seeking…but the process for us has become so much more efficient that, in fact, yes, it is eliminating jobs.”
The standard reasoning in Canada’s P&C insurance industry is that generative AI won’t replace underwriters, rather it will create efficiency or even allow you to train AI as your underwriting assistant. And while this is true — “we’re finding that every day, we develop a new tactic to reduce the load on the underwriters,” Billyard says, which is resulting in job displacement.
Much of Billyard’s presentation focused on practical applications of using GenAI in his brokerage. For example, BIG uses retrieval-augmented generation (RAG) to train a custom model on underwriting manuals.
“So, we feed our underwriting manuals in this RAG model, and then we can query it with complex questions — ‘Does Aviva cover this particular kind of risk?’ Or, ‘I have a home unprotected up north. Which markets do I want to go with this risk?’” he says.
“It saves our brokers from keyword querying and reading insurance manuals.”
Start to finish
BIG built a platform that manages the sales and underwriting process from the moment a broker starts gathering client information until the policy lands on a carrier’s portal and is bound.
Billyard uses the example of an auto policy, which has supporting documents that need to be interpreted in nuanced ways. These can include dozens of regulatory documents, commission and privacy disclosures, tax disclosures, verification of bills of sales, applying discounts, etc. GenAI is then used to strip every data point off those documents.
“This process takes about 30 seconds and runs through 130 different validations,” Billyard says. “That underwriting process used to take 20 minutes for the broker; now it takes about 20 seconds.
“So what we’ve done is build a system that guides us through this policy and underwrites the policy as it’s being written. It’s almost like having…an AI underwriter over your shoulder as you’re writing the policy.”
Billyard says this approach’s benefits include less premium leakage, reduced E&O exposure, and greater control over data than ever before.
“I can observe, among my entire broker force, which brokers are selling the lowest kilometres, which brokers are overusing winter tire discounts, which brokers are overusing pleasure use. These are all the elements of premium leakage that we weren’t ever able to control before…”
Now BIG can monitor and manage its broker workforce in real time. At the end of the process, underwriters still review each submission.
“It has resulted in us writing the best and cleanest and most accurate business that we’ve ever been able to write before,” Billyard says. “And I really feel that this is just scratching the surface of the art of the possible and the practical uses of generative AI in the brokerage business.”
Feature image by iStock.com/cherdchai chawienghong
