Marsh McLennan CIO’s Evolution: From Intern to Managing a Team of 5,000

Technology

Marsh McLennan CIO’s Evolution: From Intern to Managing a Team of 5,000

November 27, 2024

Marsh McLennan CIO’s journey: From summer job to leading a team of 5,000

Paul Beswick first joined Marsh McLennan in 1995 when he took a summer job at the insurance brokerage and management consulting firm. He’s never left and for the past five years has led a team of over 5,000 technologists as global chief information officer.

“My job has not really been to get deep into the details of what projects are being done,” says Beswick. “We have business unit CIOs who do that. They’re much better at it than I would be.”

For most of his career at Marsh McLennan, Beswick worked for the management consulting business Oliver Wyman, specializing in the retail sector. He was a consultant for more than two decades, longer than Beswick intended, because he enjoyed helping retailers implement technology solutions.

Beswick became CIO in January 2021 and since then has focused on building a culture that promotes sharing the best ideas that can solve technology needs for all four of the company’s business divisions, which include Marsh, Guy Carpenter, Mercer, and Oliver Wyman. Those businesses bring together more than 85,000 colleagues that offer risk, strategy, and talent management services, generating $23 billion in annual revenue, placing Marsh McLennan at 180 on the Fortune 500.

Each Friday morning, Beswick hosts a conversation with a colleague to talk about their career history, what work they are doing at Marsh McLennan, and what they do during their free time. The company hosts monthly tech talks centered on business themes like cybersecurity and cloud FinOps, the latter an operational framework that helps companies manage their cloud costs.

“It’s a fantastic way to get these little views into different parts of a very big business and diverse organization and personalize it,” says Beswick. “Within technology, you tend to get a lot of introverts. There’s a lot of really good stuff that happens that no one talks about.”

A few key projects that kept him busy early in his tenure as CIO included reorganizing the technology teams to create more shared services, for functions like infrastructure and cybersecurity, to run across all the various divisions.

Beswick’s thinking also evolved on cloud. The journey to the public cloud began in the middle of the 2010s and envisioned retaining six global data centers, two apiece in the U.S., Europe, and APAC regions. They’ve since dropped down to one in each market and are now working to exit data centers entirely.

The migration to cloud has no firm end date, says Beswick, who is wary of overspending and of the risk of disrupting the full enterprise if every system is updated too quickly. Amazon Web Services is his main strategic partner, but Beswick expects to be multi-cloud for the foreseeable future and works with Microsoft Azure, Google, and Oracle.

”You’re always squarely about picking one vendor—although I’d say AWS has been great—just because you feel sort of locked in,” says Beswick.

Beswick also led the development of LenAI, the company’s internally developed generative AI tool that’s used to summarize meetings, pull data from documents, and write drafts of presentations and emails. Since it rolled out roughly 15 months ago, 20 million requests have come in from employees, at a rate of about 500,000 each week. The large language models are rented, mostly from OpenAI via Microsoft Azure, but everything else is built by Marsh McLennan.

It took the company less than two days to build the first version of LenAI and Beswick says he’s happy about the experience of building the solution versus buying something off the shelf. The technology team has since gotten into a cadence of developing new capabilities that are built into the tool every few weeks.

Marsh McLennan launched a generative AI “academy” to speed up training and accelerate usage. Around 25,000 employees use LenAI each week.

“If you want to use it, great. If you don’t, that’s fine too,” says Beswick. “There’s no cost savings target. It’s just a tool that could be useful.”

John Kell

Send thoughts or suggestions to CIO Intelligence here.

NEWS PACKETS

The Qualcomm-Intel merger looks shaky. Bloomberg has reported that Qualcomm’s interest in buying rival Intel has faded, with some complications including antitrust scrutiny and $50 billion in Intel debt. Separately, the U.S. Department of Commerce awarded nearly $7.9 billion to Intel under the U.S. CHIPS and Science Act, funding that will go toward projects at Intel facilities in four states including Arizona and New Mexico. Intel has said it plans to invest around $100 billion in domestic manufacturing, including new builds and refurbishments.

Software supply-chain company hit by ransomware attack. Blue Yonder, an Arizona-based software firm acquired by Panasonic in 2021, disclosed that it was stung by a ransomware attack that affected a private cloud computing service the company provides to some customers. Blue Yonder, which counts U.S. and U.K. grocery stores and Fortune 500 firms as clients, hasn’t disclosed how many of its 3,000 corporate customers were affected by the incident. However, some retailers including Starbucks and U.K. grocery chains Sainsbury’s and Morrisons, have said they were impacted and shifted to backup systems and in some cases, manually calculating employees’ pay.

Amazon’s latest big AI bets include chips, Anthropic investment. Amazon will invest another $4 billion in the startup Anthropic, adding to the more than $8 billion that the tech giant has poured into the OpenAI rival over the past 14 months. As the New York Times reports, Anthropic is a customer of Amazon’s data centers, cloud-computing platform and semiconductors, meaning that Anthropic puts some of its resources back into Amazon as it pays for those services. Bloomberg this week published a feature that focused on efforts by Amazon’s engineers to create the first two generations of the company’s AI semiconductors, allowing it to better compete with AI chip making giant Nvidia and rely less on that company’s grip on a market worth north of $100 billion.

ADOPTION CURVE

Desire for revenue generation from gen AI outpacing productivity gains. A quarterly survey conducted by Big Four accounting firm KPMG showed that while revenue generation continues to be the top ROI metric for generative AI—hovering at around 50% for the first three quarters of the year—productivity measures have become less emphasized, slipping from 51% in the first quarter to 36% in the third.

The survey also found that while 70% of leaders have received mandatory generative AI skills training, only 28% of the broader workforce has been mandated training. That bifurcated approach could be a miss, warns Per Edin, KPMG’s U.S. AI leader and board member, saying “organizations that fail to extend training to their broader workforce risk falling behind on the adoption curve and missing out on the value opportunities.”

Courtesy of KPMG