Larry, a senior vice-president of analytics at Optum, discusses his role in the artificial intelligence space and addressing problems in the healthcare industry.

“I’ve worked with Optum, part of UnitedHealth Group, for more than 20 years in a range of roles across data and analytics, spanning various business units,” said Larry, who is currently a senior vice-president of analytics. 

Throughout his career, he has focused on a number of areas, such as learning core business processes, the financial levers that impact performance and the wider dynamics of the US healthcare market. 

He told SiliconRepublic.com, “That grounding in business and industry knowledge, paired with an understanding of how analytics can influence and drive strategy and day-to-day execution, opened up broader opportunities along the way. 

“This is what’s allowed me to progress to leading a global team of more than 700 people focused on some of the hardest problems in healthcare, making the US healthcare system work better.”

My role is focused on setting the analytics strategy that supports business priorities across multiple business units, evaluating which technologies to invest in and developing the talent to deliver on both. Alongside that, I work with my leadership team to identify where analytics can drive incremental value, both from a commercial revenue-growth perspective and through an internal cost and productivity lens. 

Year to year, that means making sure the capability, the talent and the platforms are in place to meet the needs of the business and working with our technology partners to look ahead at where the technology is going, so our strategy is aligned to where capabilities are heading, not where they have been. Day to day, it’s working with key stakeholders to prioritise initiatives and adjust as the business shifts. 

As an enterprise, we are currently embracing an AI paradigm across our businesses, and scaling AI is critical to our future success. Getting there depends on two things.

First: not simply using AI to automate existing processes, but reimagining what those processes could be given what advanced technologies can now do. Second: applying the right AI capability to the right problem.

That means resisting the temptation to throw the latest frontier model at everything; often, a traditional machine learning model is better suited and considerably more cost-effective. Key to all of this is trusted data and responsible governance. In healthcare, we have to earn the right to scale AI, protecting privacy, being clear about where human judgement remains essential, and monitoring for unintended bias or variation in outcomes. 

Five things come up consistently.

First, testing and maintaining accuracy, building the evaluation discipline to know whether a model is performing and keeping it performing over time. Second, legal and compliance review, which in healthcare is rightly rigorous and needs to be designed into the process rather than as an afterthought.

Third, stakeholder adoption – technology is often the easy part, changing how people work is the hard part. Fourth, integration into existing workflows and systems is oftentimes the longest pole in the tent. And finally, proving value consistently. A technically successful pilot is not enough. We need to measure adoption, operational impact and ultimately, whether it improves outcomes for the people and organisations we serve. 

Give people headspace to innovate and to keep developing their skills with the latest technologies and approaches. Provide standard frameworks and playbooks for testing and scaling, so teams aren’t reinventing the path to production every time. And put the support structures around them to manage that at scale that include clear routes through compliance, review and deployment.

Leaders also need to create psychological safety; teams must be able to surface risks, challenge assumptions and stop or redesign a solution when the evidence says they should. 

Across teams, it starts with understanding the business and the customer need, then designing something that addresses it. That takes a mix of people – some who know the business, some who know the technology and an operating model that cleanly connects the two. You also need strong data skills, with a deep understanding of the data, where it came from and how good it is, as well as data governance and data engineering as critical skills for any new AI project.  

For my role, it is about making sure those resources are available and that the connections are being made. It’s also about asking the challenging questions – are we solving the right problem, and is this the right capability and resources to solve it with? 

The biggest shift has been agentic AI moving from demo to production, and AI evolving from a productivity tool into a core component of the operating model through AI-enabled workflows, decision support and end-to-end process orchestration.

Twelve months ago, the discussion centred on what these technologies might do. Today, the focus is on what they have delivered, how we measure and govern performance, and where we scale next.  

For Optum, this has opened a broader conversation about becoming AI-native across our business processes. The opportunity is no longer simply automating individual tasks, but redesigning how work gets done – simplifying operations, improving affordability, increasing speed and quality, and ultimately delivering a better healthcare experience for patients and care providers. 

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.