While much of the industry debate on AI has centred on underwriting and distribution, Greg Smith (pictured) believes the front lines of claims are actually poised to benefit from automation – without losing the human element policyholders depend on most.
The president of Crawford & Company (Canada) expects 2026 to bring more economic pressure and continued catastrophe risk, but he argues the fundamental demands on adjusters will not change: rigorous investigation, clear communication and sound judgment, supported – not displaced – by technology.
Smith is well aware of the hype around generative tools and automation in insurance. But when he thinks about what will matter this year for Canadian claims teams, he starts with basics rather than code.
He said adjusters will still need to carry out “disciplined investigations that leverage data and analytics,” while also drawing on their own experience to spot opportunistic fraud and be “sensitive to some of the hardship that people are experiencing, both related to the loss and just related to the general condition of the economy.”
On top of that, he sees a dual responsibility: moving quickly to validate and pay legitimate claims, and standing up defensible decisions when coverage is not triggered. That means the ability not only to make tough calls, but to articulate them.
Adjusters, he said, must be able to “explain it really clearly so the policyholder understands” and document files so that the insurer can defend its position “if it’s challenged by the policyholder at some point in the future.”
In Smith’s view, those human capabilities – investigation, empathy, explanation – are what distinguish effective claims work. The promise of technology is to create more space for them.
Smith rejects the idea that AI is poised to “exterminate” the human side of claims. Instead, he sees near‑term tools as a way to strip out tasks that get between adjusters and customers.
He pointed to the first notice of loss as a clear example. In a traditional call centre workflow, an operator speaking to someone who just had a car accident or house fire is forced to split focus between the conversation and a screen full of fields.
“You’re trying to listen to the policyholder and understand what happened,” he said, but also chasing postal codes and phone numbers and “trying to fill out the first screen before you get to the second screen.”
With AI listening in and populating systems in the background, Smith said that operator should be able to “push that keyboard to the side and really focus in on the conversation.” Whether the caller mentions their accident before their address becomes irrelevant if the tool can interpret the story, look up the routing details and line up the nearest adjuster or contractor.
Those functions – assigning resources, validating contact details, filling databases – are critical, but they do not need to crowd out the human interaction. Smith sees a future where AI “does all the work that the adjuster currently is doing or used to do manually,” freeing up time for what he calls the “important” parts: listening, explaining next steps, and applying technical judgment.
“There will always be a need for human empathy and human support for the most severe losses,” he said, particularly where someone has “experienced a trauma” or a significant loss. If anything, he sees automation strengthening those connections because adjusters spend less time typing and more time talking.
Beyond empathy and communication, Smith links effective use of technology to basic operational rigour. The same efficiencies that help on everyday claims become critical when severity, frequency or both spike, he argued.
He said that as losses become larger or more numerous, “small details matter” in how work is organised – from triage and assignment to the way tools and systems support field staff. Minor inefficiencies can “really amplify and multiply themselves in a cat situation,” which is why Crawford is using quieter periods to refine its processes.
The firm is investing time now, he said, to “iron out the wrinkles, to lock down those processes so that we’re operating at peak efficiency when it matters most and when it really is impactful on policyholder experience.”
