How AI Can Connect Pharma Manufacturing Data Systems

In part 2 of a 3-part interview, Susan Schniepp, Regulatory Compliance Associates Inc., and PDA representative, during PDA Week 2026, discusses one of the most pressing challenges facing modern drug manufacturing: understanding how the many interconnected data systems within a facility work together to protect product quality.

At the core of Schniepp's argument is the idea that the industry has long treated critical data streams, smoke studies, particulate matter results, environmental monitoring, as isolated data points rather than as parts of a unified picture. She sees this siloed thinking as a significant vulnerability and believes emerging technology may hold the answer. As she explains, "I think one of the things that's different now moving forward for a lot of these initiatives, including our Annex 1 contamination control strategy, is what people are not getting or haven't been getting — and this is maybe where AI can actually help us — is how all of these pieces fall into place to protect the product."

To illustrate the risk of incremental, unexamined deviations, Schniepp draws a vivid analogy from her time in chemistry labs, where small individual tweaks to a validated method could collectively push it out of compliance. She applies the same logic to manufacturing environments. If deviations in environmental monitoring, particulate matter, and water results are each within tolerance individually, at what point does their combination render the system non-validated?

Schniepp also highlights a unique operational reality for parenteral and injectable manufacturers. Unlike other facilities, these sterile suites cannot simply be shut down over a weekend. "The parenteral suite — that's why you see so many contract manufacturers — because they can keep those lines running and maintained, and they're productive because they're 24/7 running."

Schniepp calls for the industry to embrace a more holistic view of quality data, one where background systems and batch snapshots are understood together, not in isolation.