As reinsurance portfolios become more complex and accumulation risk harder to detect, artificial intelligence is beginning to shift from conceptual promise to operational tool. For reinsurers, the challenge is not whether to adopt AI, but how to apply it in a way that strengthens underwriting discipline and portfolio control.
At Swiss Re, AI is increasingly being used to improve how underwriters access data, identify trends, and gain earlier visibility into emerging risks, according to Kera McDonald (pictured), group chief underwriting officer.
As portfolios grow more interconnected, the risk for reinsurers is no longer just mispricing individual risks, but failing to see how exposures aggregate until it is too late.
“Taking risk is our job,” McDonald said. “AI helps us understand that risk better and faster, but it does not make decisions for us.”
McDonald, who has spent more than two decades at Swiss Re in underwriting, actuarial, and risk management roles, says the focus has been on practical applications that support judgment rather than replace it, particularly as volatility and portfolio complexity increase.
For McDonald, AI’s most tangible contribution today is not advanced modelling, but the removal of long-standing friction in underwriting workflows, particularly around data.
“We have a lot of data in this industry,” she said. “The challenge has always been unlocking it in a consistent and usable way.”
Data ingestion has been the most immediate win. Where underwriters once relied on fragmented processes to extract insight from unstructured information, AI now enables faster, more standardised intake, freeing underwriters to focus on judgment rather than administration.
“That is low-hanging fruit, but it matters,” McDonald said. “It allows underwriters to spend time on the decisions, which is where they add the most value.”
Beyond ingestion, AI is improving access to information Swiss Re already holds but previously struggled to analyse at scale. Contract batch-screening, natural-language search, and portfolio-wide trend analysis are already changing how risk is reviewed.
“We had the data,” she said. “What has changed is how easy it is to see what is actually inside our contracts and identify trends across the portfolio.”
McDonald is clear that AI is not a shortcut to higher prices in competitive markets. Instead, its value lies in sharper risk selection and stronger governance.
“Market dynamics will always push pricing to a place where underwriting companies hopefully earn slightly above their cost of capital,” McDonald said, adding that improved risk understanding is critical in competitive markets.
That understanding supports portfolio profitability by improving consistency and visibility. AI-enabled processes allow Swiss Re to monitor how risks enter portfolios, how underwriting practices evolve, and where discipline may be slipping.
“It is an enabler, but it is also a guardrail,” McDonald said. “As long as we treat it appropriately, it helps reinforce underwriting standards rather than dilute them.”
For reinsurers without that visibility, portfolio discipline can erode quietly, not through individual underwriting failures, but through blind spots that only become visible after loss events.
As volatility increases and risk correlations become harder to anticipate, McDonald sees AI’s greatest value in identifying accumulations that would otherwise remain hidden.
“For standard risks that are already well controlled, humans will continue to manage those,” she said. “Where AI helps is in finding new trends or accumulations buried deep within the portfolio.”
By screening large volumes of contracts simultaneously, AI can surface emerging concentrations earlier, giving portfolio owners more time to respond.
“The real question is whether we can react faster. That is where AI adds the most,” McDonald said.
Swiss Re draws a firm line around decision-making authority. AI supports underwriting decisions, but it does not replace them.
“We talk about ‘augmented underwriting,’” McDonald said. “The human remains in the loop, especially given the nuance and complexity of reinsurance.”
That approach is underpinned by years of investment in data infrastructure – creating unified datasets with common definitions and governance. AI builds on that foundation rather than bypassing it.
“This remains a people business,” she said. “The tools change how the work looks, but not what underwriting is at its core.”
McDonald expects data literacy and comfort with AI tools to become essential underwriting skills. What will not change is the need for critical judgment. “The act of underwriting will remain,” she said. “The value-add in judgment will still be there.”
The greater risk, McDonald suggests, is not that AI will make the wrong decision, but that underwriters may trust it too readily during the transition phase, which could lead to mistakes.
“If underwriters believe AI has read the contract for them and stop reading it themselves, that is a problem,” she said. “Our responsibility is to design these tools in a way that controls those risks.”
AI development at Swiss Re is moving fast through focused “reimagination” exercises that rethink processes from the ground up. In roughly 10-12 weeks, the organisation can move from early reimagination to actively building a minimum viable product, with full MVP deployment typically following within six months.
Crucially, underwriters are directly involved. “You cannot automate tasks without the people who do those tasks today,” McDonald said. “If we want these tools to be useful, underwriters have to help build them.”
The response from underwriting teams has been positive, driven by curiosity and a clear link between AI tools and real underwriting challenges.
“If AI were introduced purely for the sake of technology, it would not work,” she said. “Once you address real business needs, people are very willing to engage.”
In that environment, AI is less about underwriting faster, and more about avoiding the blind spots that complexity and volatility increasingly create.
