The biggest challenge facing the biopharma industry isn’t scientific innovation. It’s manufacturing execution.

That’s the opinion of Nicolas Pivet, vice president and general manager of manufacturing and digital solutions at Cytiva. Major scientific advances have enabled the development of truly amazing therapeutics such as RNA technologies and cell and gene interventions, yet manufacturing innovation hasn’t kept pace. “Our ability to manufacture these therapies reliably, repeatedly, and at scale remains a major constraint,” Pivet tells GEN.

“The industry often focuses on symptoms [of biomanufacturing bottlenecks] such as paper batch records, disconnected laboratory information management systems, fragmented automation architectures, isolated enterprise resource planning deployments, or data silos. Those challenges are real, but they are not the root cause,” Pivet stresses.

“The deeper issue,” he says, “is that many manufacturing environments evolved over decades and were never designed as connected systems. Digital capabilities were often layered on to solve local problems rather than integrated around end-to-end manufacturing performance. The result is an environment that is increasingly data-rich but…insight-poor.”

In that environment, organizations can’t extract actionable insights early enough to affect outcomes. They have a visibility problem.

Focus on observability

The first step in resolving the bottlenecks is to solve specific problems.

“You cannot control what you cannot see,” Pivet says. He advises industry to focus first on improving observability. Gaining insights into process deviations early enables earlier intervention, which creates better decisions and improves confidence in those decisions, he explains. This is the foundation of advanced manufacturing.

“Then connect the data, workflows, and decisions required to improve the outcome,” Pivet advises. “Biopharma does not lack information. It lacks accessible, contextualized, and trusted information,” he says. “Connecting process, equipment, analytical, and workflow data establishes the foundation required for predictive analytics, advanced automation, digital twins, and AI.”

“Only after that [has occurred] should advanced control or AI-enabled decision-making become part of the discussion.”

Therefore, rather than starting with AI use cases—as many try—organizations should first focus on observability, and then address interconnectivity, data integrity, and maintaining operational context for those data flows. When those are resolved, AI can be added to the mix. “AI does not fix weak foundations,” he says. “It exposes them.”

Face organizational barriers

The barriers to digital transformation are more often organizational than technical. That’s because biomanufacturers consider change riskier than maintaining current practices. In 2025, “more than a third of biopharma companies were still in early exploration or had not started their digital transformation journey,” he says, underscoring his point.

Trust is typically the rate-limiting step, and that is established only by reliable performance. Therefore, “the organizations moving fastest are those demonstrating value on real manufacturing problems, validating outcomes, and gradually scaling successful approaches rather than attempting sweeping transformations,” Pivet observes.

Manufacturing equipment in the future will not just be smarter; it will be designed from the outset as elements of connected, interoperable systems that provide early visibility and easy integration into automated and connected workflows.

Prepare the workforce

“The conversation around AI often focuses on technology. The more important conversation is about people,” Pivet emphasizes.

Automation, agentic and generative AI, and other smart manufacturing technologies are changing roles throughout the workforce at, seemingly, warp speed. Everyone, from equipment operators, process engineers, data scientists, and quality professionals up through the executive suite, is coming to grips with new ways of working that require new skills and new ways of thinking.

“Data literacy, systems thinking, automation fluency, model interpretation, and cross-functional collaboration are becoming increasingly important,” Pivet points out. Consequently, people are spending less time collecting information and reacting to events, and more time interpreting data and optimizing processes.

The workforce, however, is still in transition. “The Global Biopharma Index reported that 29% of organizations experience severe shortages in digital and AI-related skills,” he says.

“Workforce readiness is one of the defining challenges of this decade,” Pivet continues. “The most successful manufacturers I encounter are not necessarily the ones deploying the most advanced technologies. They are the organizations creating confidence around new ways of working.”

Within five years, he predicts a plethora of hybrid roles that blend both process and digital expertise. Future titles may include digital process engineer, manufacturing data scientist, process intelligence specialist, and manufacturing systems architect, he adds.

Filling those new roles means preparing the workforce today. He advocates upskilling individuals enterprise-wide, ensuring the various functional teams learn together, rather than treating digital transformation as the purview of any particular group.

Even today, across industries, traditional learning programs are being complemented by immersive learning experiences. Biopharma will be no different, especially given its growing complexity and limited opportunities for hands-on learning in the face of good manufacturing requirements, production schedules, and equipment availability.

Immersive learning experiences involving mixed reality and augmented reality “allow operators, technicians, and engineers to interact with equipment and workflows, and troubleshooting scenarios in highly realistic digital environments,” he says. By accelerating learning, enhancing retention, and building confidence, workers are better prepared when they enter live manufacturing environments.

Eventually, Pivet says, “Immersive learning may become as important to workforce development as flight simulators became to aviation training.”

Ease tech transfer

The tech transfer process also has a major bottleneck, but it has nothing to do with technology or human factors.

“The most common reason transfers fail is not because process instructions are missing,” he says. “It is because process understanding fails to transfer with them. Receiving organizations often inherit process descriptions, specifications, and operating procedures, but not the context explaining why those parameters matter or how the process behaves under changing conditions.”

Those gaps are becoming ever more important as therapies become increasingly complex. Successful tech transfer, therefore, includes transferring process understanding. Digital twins help minimize the gap between process instructions and process understanding by enabling teams “to test assumptions virtually before testing them physically,” Pivet says, thus enhancing success. In this way, “…uncertainty is removed rather than relocated.”

Lead with digital change

If you can prioritize only one area for improvement, Pivet advises choosing digital infrastructure because, he explains, “people and processes increasingly depend on information. Digital infrastructure creates the conditions for successful technology transfer, effective workforce development, predictive analytics, process understanding, and operational improvement. Without trusted information flows, organizations struggle to scale knowledge, replicate success, or improve decision quality.”

High-quality information flows, he predicts, will be a major determinant of manufacturing competitiveness in the coming decade. “The companies best positioned to lead the next era of biomanufacturing will be those that combine AI initiatives with strong connected-data and observability foundations,” he suggests. By doing so, they will be better able to translate scientific innovation into predictable, scalable, operational performance.

“Biopharma does not have an imagination problem,” Pivet reiterates. “It has an execution problem. Closing that execution gap is one of the greatest opportunities our industry has to accelerate patient access, improve resilience, and unlock the full value of the science being developed today.”