Quality carries a significant operational burden in MedTech. Regulatory requirements, documentation, investigations and audits demand time from highly skilled teams, while global supply chain complexity adds further pressure. As that environment expands, the industry is beginning to ask whether AI can change how quality work is performed.

These challenges shaped an AdvaMed webinar moderated by Partha Marella, EVP of Life Sciences at HCLTech, featuring Noor Malki, Senior Vice President and Head of Regulatory Affairs and Quality for Varian within Siemens Healthineers’ cancer care business and Keith Lavery, Vice President of Data Analytics and Artificial Intelligence at Zimmer Biomet. 

The conversation highlighted that quality costs have remained difficult to shift despite years of process improvement.

“What really has changed is the arrival of Agentic AI systems that don't just assist but can actually take some action and change the course of quality and systems,” said Marella. 

Complexity keeps quality costs high

For Malki, the pressure starts with the environment around quality and regulatory teams. Organizations are managing growing volumes of regulations, standards, data and documentation. That administrative burden uses capacity that could be directed toward managing risk and preventing problems earlier.

“In my view, the biggest driver of cost is complexity,” she said. 

The most time-consuming areas include investigations, complaints, product issues and audits. Complaint handling shows why. Teams may need to gather information from multiple sources, connect related records, review evidence, reproduce product behavior, identify root causes and prepare conclusions.

AI can reduce the effort required to assemble and organize that information. Specialists can then spend more time applying judgment and deciding what action is needed to prevent recurrence.

Lavery sees similar potential in complaint workflows and post-market surveillance, where classification, evidence gathering and evaluation remain highly manual. These activities provide natural points for AI to take on more of the cognitive workload. 

Agentic AI expands the scope of automation

Agentic AI extends that opportunity because the technology can begin to make decisions and act across systems. Greater autonomy also increases the importance of understanding how those decisions are governed.

“It's where and when the agents are actually making decisions or have the ability to poke into different systems or activities and make the decisions. That's the differentiator from a traditional automation flow,” said Lavery. 

The technology can move between applications, retrieve information and determine whether it needs additional data to complete a task. As those capabilities develop, quality teams will need visibility into the decisions agents make and the controls around them.

Malki described opportunities across pre-market regulatory work, post-market processes, complaints and audits. Within her organization, AI is also supporting different phases of quality workflows and helping users complete tasks that previously required more manual support. 

This can create capacity for specialists to apply judgment where it matters most.

“The way we're thinking about it is human-led, then AI-powered,” she said. 

Human oversight remains central

Malki explained that existing review points remain in place within deployed AI-supported quality processes. An individual still checks the output before the process moves forward.

Lavery described a similar approach. Some decisions retain traditional human checks, while others are reviewed through analytics showing how AI classified information and reached a conclusion. Teams can then identify where training data or the model may require adjustment. 

Marella expects human contribution to grow as these systems mature.

“The human element is going to play a bigger role rather than a diminished role,” he said. 

Quality teams may increasingly spend their time evaluating AI decisions, managing exceptions and defining the boundaries within which agents operate. Lavery also highlighted cybersecurity as a growing concern as organizations consider the quality and integrity of decisions made by machines. 

Scaling requires process redesign

Moving from a pilot into enterprise deployment depends on whether the business is prepared to change the underlying workflow.

“What separates [pilots from enterprise deployment] is a business team that's willing to invest the time to redevelop their process or their way of doing things,” said Lavery.

AI can improve individual steps within an existing workflow, but greater value comes when teams reconsider the process itself. Lavery said pilots are less likely to deliver the expected return when organizations layer AI onto existing ways of working without addressing unnecessary steps, systems or handoffs.

Malki also emphasized the people involved in scaling AI. Enterprise adoption requires alignment across stakeholders, while employees may have concerns about job security and changing responsibilities. Leaders need to explain how the technology will support people's work, provide training and recognize that teams will adapt at different speeds. 

Measure value through the work

The discussion repeatedly returned to measurable value. Quality organizations can assess AI against the cost of risk, changes in workload and improvements in cycle times. Lavery described a continual pipeline of use cases, with teams measuring whether AI can move a relevant KPI and redirecting effort when it cannot. 

Malki's view of success reaches across the product lifecycle, including shorter investigations, faster submissions and greater consistency in risk assessment.

“Success in my view is achieving scale and creating end-to-end visibility,” she said. 

The broader purpose remains patient outcomes. Lavery argued that greater efficiency can help organizations get products to market faster while maintaining quality, while Marella linked that performance back to trust with patients and the wider healthcare ecosystem.

That is where the value of Agentic AI in MedTech quality will ultimately be judged. It can reduce administrative effort, help teams identify issues earlier and create more capacity for judgment. Real impact will depend on whether organizations redesign processes around the technology and build the governance and workforce confidence needed to use it well.

The future ROI of Agentic AI in MedTech quality

The shape of an Agentic AI-enabled quality organization will continue to evolve, but patient safety will remain central to how the technology is applied.

As more administrative and process-intensive work is supported by AI, quality and regulatory teams can spend more time on risk prevention and the decisions that require human judgment. Over time, that could strengthen their role as strategic partners to the business, with value measured through better outcomes, greater efficiency and continued trust.