Artificial intelligence is increasingly being used in insurance catastrophe response to support claims triage, back-office operations and workforce scaling during surge events. For Deloitte Canada’s Colin Asselstine (pictured right) and Chris Duvinage (pictured left), the most immediate gains are practical rather than transformative: smarter triage, improved operational support and faster deployment of staff during CAT events.
Asselstine, Deloitte Canada’s insurance claims leader, says the question for most carriers is no longer whether to use AI, but where it actually adds value.
“In a CAT scenario, the first thing that you need to do is triage your claims in a way that helps you to organize them,” he told Insurance Business. “These are ones that are VAC or food spoilage – they’re simple, they’re low risk. How do I get those ones settled as quickly as possible, so my adjusters are dealing with individuals that have meaningful impacts, and how do I get them back on their feet?”
Historically, adjusters often had to set files aside and revisit them days later. AI‑enabled tools can now scan incoming claims and push the most urgent cases to the top of the queue.
“You can take all of your information and figure out where you need to respond in a timely manner, to save costs and improve your customer experience,” Asselstine said. “From a CAT perspective, that’s where carriers should go first – getting quickly to the customers that need help first.”
Beyond triage, a lot of current AI use is about making adjusters and call‑centre staff more effective.
“The other thing folks are looking at is just having people in the back office get assistance from AI,” said Duvinage, Deloitte Canada’s national property & casualty insurance segment leader. “Listening to calls, transcribing calls, suggesting next‑best action, and honestly just driving down the call time.”
“For instance, instead of a call taking 25 minutes, it’s now 15,” he says. “That’s huge, given the number of calls they get, especially during a CAT.”
AI prompts can also flag customers who haven’t been followed up with, highlight missing information and keep notes more accurate across multiple systems. For adjusters and vendors dealing with different insurer platforms and question sets, that matters.
It also changes the feel of the interaction. With transcription and guidance happening in the background, staff can focus more on the conversation and less on rigid screen flows.
“It’s allowing you to be a lot more natural on the phone and deal empathetically with a customer, versus scripting the way your conversation goes based on the screen layout,” Duvinage says.
The other pressure point AI is starting to address is human capacity when CAT events hit.
“When you talk about a CAT event, the biggest opportunity is how you scale,” Asselstine says. “How do you get more people on the phone to pick up those customers as they’re coming in?”
Traditionally, carriers leaned on property teams, then pulled in adjusters from auto. Those staff often had only partial familiarity with property coverages and processes.
“Insurance companies are saying, ‘Okay, I’m using my property department now. Where can I get more adjusters? I’m going to go to auto. They don’t have the skills, or they have a portion of them, so how do I supplement?’” he explains.
AI‑driven guidance tools can quickly query SOPs, look up coverage and suggest the next question, letting staff from auto, injury and accident benefits support property calls without months of training.
“You can scale a lot faster with the right information, ask the right questions, and that’s allowing you to cut down your hold times quite significantly,” Asselstine says. “That’s improving customer service – and it’s happening right now, because it’s not that risky.”
Some carriers are already baking this into their CAT playbooks, pre‑training broader groups of employees and using AI as a real‑time co‑pilot once a surge threshold is reached.
For all the back‑office progress, fully autonomous, customer‑facing AI in claims is still at an early stage.
“Has anybody launched anything at scale yet to the customer? Customer‑facing? No,” Duvinage says. “It’s still very much digital intake. It’s not like an AI agent that talks to you.”
He does expect that to change.
“You can imagine a world where these AI agents on the phone get so good you honestly can’t really tell if it’s a human or not,” he says. “What used to be pretty clunky is now getting to a point where you can barely tell the difference. Eventually, these AI agents will be better trained, instantly trained, instantly scalable.”
In that world, he sees clear segmentation: some policyholders will always wait to speak to a person, others will happily “skip the line” for a virtual agent that delivers fast, accurate answers in a CAT.
Any move towards more automated decision‑making will attract regulatory attention, and insurers are already working on internal controls.
“As an industry, we cannot afford not to figure this out at pace, because customers are expecting it,” Duvinage says. “Customers are getting that type of service from other sectors, in financial services, consumer and so on.”
Asselstine says much of Deloitte’s work with carriers is now about designing governance frameworks.
“A lot of the work we’re doing right now is advising on how to structure and put proper controls in place,” he explains. “You want to be able to monitor your customer experience, monitor the risk and the payouts, and put in proper stop‑gaps that make human intervention where you want comfort.”
That means flags when a model is about to make a higher‑impact claims decision, rules that force a hand‑off to a human in certain scenarios, and continuous monitoring of outcomes.
“We’re just at the beginning step of that journey,” Asselstine says. “The question is how do you jump in, but in a way that ensures safety for customers and for the loss ratio companies are trying to protect as well.”
