Canada's largest property-and-casualty insurer says AI earns it C$250 million a year. Two big life insurers gave no dollar figure

Most companies describe what AI might save them someday. Intact Financial, Canada's largest property-and-casualty insurer, put a number on what it says AI already earns. On its third-quarter call on November 4, CEO Charles Brindamour said AI now produces close to C$250 million a year in recurring pre-tax earnings. He said the company expects to double that within 24 months.

That kind of disclosure is still rare, and this week showed how rare. Manulife and Sun Life, two of Canada's largest life insurers, released results that evening. Both described AI tools spreading through advisor, underwriting and claims teams. Neither release put a figure on what those tools add to profit.

For investors, the difference matters. A dollar figure tied to a stated method can be tracked each quarter and challenged. A list of rollouts cannot. Intact beat analysts' per-share profit forecast by nearly 10% and its shares rose more than 6% after the results. How much of that move came from the AI target, rather than the earnings beat, is impossible to separate.

What Intact is claiming, and how it says it measures it

Intact's AI work is concentrated in pricing, the part of an insurer's business that decides which customers to take and what to charge them. The company says it has more than 600 proprietary models in production across Canada, the US, and the UK and Ireland. Machine learning now sets prices for 94% of its Canadian personal lines and 75% of its commercial lines.

Management said about 60% of the next C$250 million should come from a lower loss ratio, meaning less of each premium dollar paid out in claims. The rest should come from lower operating costs. The company also cited operational gains: a 70% cut in turnaround time for commercial quotes sent by brokers, and about half the preparation time for legal claim documents.

The measurement method is the most useful detail for other executives. Brindamour said Intact compares results in areas using AI against a control sample, a group run without the new models. He argued the figure is real money, not a projection that higher spending elsewhere will quietly erase. He also said the outperformance seen where AI is used most is larger than the amount the company counts.

What the number does not show

The figure is company-reported and unaudited. Intact did not publish how its control samples are built, which costs are netted out, or how the total has changed over time. Outsiders cannot yet check it.

AI is also not yet lowering Intact's overall costs. Chief financial officer Ken Anderson said the expense ratio, operating costs as a share of premiums, will land at the high end of its 33% to 34% range in 2026, as the company keeps investing in data, AI and technology. Chief operating officer Patrick Barbeau told analysts some of the AI gains will fund further AI investment rather than flow straight to profit.

And the quarter itself was weak. Net operating income per share fell 19% to C$3.63. Catastrophe losses of about C$660 million pushed the combined ratio, claims plus expenses as a share of premiums, to 94.0% from 89.8% a year earlier. Better pricing did not offset a season of severe weather.

Scale matters too. Intact earned C$641 million in net operating income for common shareholders in the third quarter alone. A C$250 million annual AI contribution is meaningful, but it is not yet transforming the business.

The data moat argument, and the threat Intact concedes

Brindamour's larger pitch was that general-purpose AI will widen the gap between large and small insurers, not close it. "You can't prompt your way to a sophisticated pricing model," he said. Intact says it is twice the size of its nearest competitor, and argues that scale gives it a large data advantage. It handles more than 95% of claims in-house, which feeds its models. The company has been building in-house AI since founding its Intact Lab more than 12 years ago.

He did concede one risk. A UBS analyst asked whether AI shopping agents, software that compares and buys policies for consumers, could cut insurers off from customers. Brindamour said parts of the broker channel could come under pressure, especially brokers without scale. He argued Intact's brands, direct digital channels and pricing skill leave it well placed.

Its life-insurance peers reported AI progress in operational terms instead. Manulife expanded its partnership with Microsoft through a new five-year agreement. It is extending Microsoft 365 Copilot to more than 30,000 employees and deploying Microsoft's Agent 365 to strengthen oversight of its AI use. Its US Quick Quote tool now automates more than 70% of preliminary underwriting assessments, up from nearly 50% in the first quarter. An AI tool rolled out to its wholesalers saves about 30 minutes per session by automating notes, Manulife said.

Sun Life launched an AI productivity assistant for advisors across Asia. It said automated adjudication of medical claims in Indonesia raised straight-through processing, claims settled with no human touch, by nearly 20% in two months.

Those are real gains, but none comes with a profit figure. For boards and buyers, Intact's approach offers a template worth copying: pick a core profit lever, run AI against a control group and report the difference in dollars. Its caveats offer a second lesson. Until a company shows its method, a measured AI number is still a claim.