For companies whose business units have traditionally operated independently, centralizing IT operations under one strategy can reap significant benefits — especially when it comes to offering a holistic customer experience and establishing a unified data foundation for leveraging the latest emerging technologies.
That’s where EVP and CIO Kathy Kay found herself in coming to Principal Financial Group from PG&E in May 2020 with a desire to lead an aggressive plan to adopt digital technologies, ranging from the cloud to AI.
To that point, IT pros within the company’s three US business units, its international business unit, and its holding company had implemented cloud computing and other technologies ad hoc. As such, Kay’s biggest challenge in setting the stage for her digital ambitions — and one that has involved the entire C-suite — has been unifying the IT operations of Principal Financial’s individual businesses units in order to develop an enterprise-wide data foundation and cloud architecture on which to build its next-generation services.
The Fortune 500 company, born an insurer in Des Moines, Iowa, roughly a decade after the Civil War ended, is under pressure to provide customers with an integrated experience, particularly due to its expanded financial services portfolio, including the acquisition of Wells Fargo’s Institutional Retirement and Trust (IRT) business, Kay says.
“What caused us to move much faster to the cloud was aligning our technology assets with the capabilities we needed to enable a modern business strategy,” she says. “The customer had a 401(k) with us, for example, insurance benefits, and also asset management services, and [each] of those units did not know they shared a customer. Customers expect to be treated holistically.”
The multinational operates in 80 countries, has a market capitalization of $18 billion, and employs roughly 20,000. It has traditionally served SMBs and government-based retirement plans from its Iowa and Charlotte hubs, and it has sizable footprints in Chile, Brazil, Hong Kong, Malaysia, Singapore, and India.
To tackle the unification strategy, Kay and her C-suite colleagues established Enterprise Business Solutions, an internal technology division focused on modernization. IDC analyst Thomas Shuster notes that moves like these better position Principal Financial to compete against major financial services firms.
“Principal’s move to integrate their asset management and pension businesses makes a lot of sense given they’re highly interrelated and it opens the door for a better and differentiated client experience. The proof is in the implementation and replication of successful collaboration,” says Shuster, adding that the acquisition of Wells Fargo’s IRT business has given Principal relatively rapid access to market share.
“Their creation of the internal technology group dovetails well with the reorganization and the business acquisition — now is a good time to get enterprise data and technology strategy right,” he says.
Laying the foundation for AI
According to Kay, the creation of an integrated enterprise data foundation across all five business units enables the company not only to drive new revenue but also to exploit new technologies.
“We had to make [that] investment to leverage AI and machine learning models for different analytical capabilities across the entire company,” Kay says.
Principal, which Kay describes as “a huge Salesforce shop,” is midway through its cloud adoption and has automated business processes using Automation Anywhere. It has also implemented chatbots, developed machine learning models using Databricks, and is dipping its toes into generative AI.
Among the machine learning models Principal uses in production for its asset management operations are propensity models, an opportunity scoring advisor, sales scoring, retention models, and outbound call scoring, as well as some quantitative investing and next-best actions, Kay says.
Principal also maintains a partnership with Evolution IQ, which “helps Principal improve the customer experience for all our disability insurance lines by substantially reducing steps and timing for claims,” says Kay, who previously served in leadership roles at GM, Comerica, and SunTrust.
Embracing gen AI mid-migration
Like many financial services companies, Principal is also investigating how gen AI can benefit its various lines of business. To this end, the company is “playing” with Microsoft’s Azure OpenAI Service, Kay says, with one generative AI application already in production.
“As the marketing teams are writing external-facing content, we used to manually review all of it from a compliance perspective,” Kay says of the common process of responding to Request for Proposals. “We have a model now that will ingest the data and then suggest better ways of communicating things in a more complaint way.”
As a backup, Principal has employees complete a compliance review of the final product but the gen AI model “is saving a lot of time,” Kay says.
To enable more AI, Principal’s engineers are using GitHub Copilot for generating code, Kay adds. After conducting an internal study of 250 engineers, the company acquired an enterprise license to open the platform up to maximize innovation and jumpstart the creation of new machine learning models and generative AI models.
Principal Financial has not completed its cloud deployment but that does not prohibit the company — or any company — from running generative AI models, IDC analyst Dave McCarthy points out.
“While it might seem logical that you would need to be in a public cloud to take full advantage of gen AI, privacy and cost concerns are driving interest in hybrid approaches,” McCarthy says, noting that the largest AI models will require the public cloud due to resources needed but not all corporate AI models need it.
“As this technology is still in its infancy, CIOs are taking the safe approach by keeping sensitive data on-premises,” McCarthy points out. “Vendors like Dell, HPE, and VMware are catering to this need by developing datacenter infrastructure that is tailored for gen AI workloads.”
