Pulmovant

Building Quality into Every Stage of Production

The tension between speed and quality is real, but I try not to frame it as a trade-off. Patients are waiting on new therapies, and a quality failure is the fastest way to delay them. Consistency starts with making the standard non-negotiable and the path to meeting it as easy as possible.

In practice, that means three things. First, build quality in at the design stage. Clear specifications, well-characterized processes and validated analytical methods so you are not trying to inspect quality into a batch at the end. Second, hold your partners to the same bar you hold yourself. In pharmaceutical chemistry, so much of the work happens across API manufacturers, CDMOs and testing labs, so strong quality agreements, supplier qualification, and regular oversight are what keep a standard consistent across sites. Third, communicate early. If operations knows what quality needs and why, you will spend far less time negotiating at the eleventh hour.

Embedding Risk Assessment into Process Development

The most valuable risk work happens before anything goes wrong. I use the principles in ICH Q9 as a backbone: identify hazards, assess severity, probability and detectability, and then decide what controls are proportionate. The discipline is in applying it early, during process development, tech transfer and change control, rather than treating it as paperwork after the fact.

Beyond formal tools, I rely on trending. Small signals, such as a drift in an impurity profile, a rise in minor deviations at one supplier or a pattern of repeat investigations, usually appear well before a failure does. I also encourage a culture where anyone can raise a concern without friction. Some of the best risk insights have come from an analyst or an operator who noticed something that did not look right.

Building Efficiency into Quality Management

I do not think compliance and efficiency are opposites. Most inefficiency in quality comes from duplicated effort, unclear ownership and processes that grew by accretion. When we simplify, we tend to improve compliance at the same time.

The starting point is a risk-based approach consistent with cGMP expectations and ICH Q10. Focus effort where the risk to the patient is highest, and do not apply the same level of scrutiny to everything.

Every decision in quality ultimately comes back to the patient on the other end.

Next, make records right the first time. Clear procedures, intuitive forms and good data integrity practices reduce errors, rework and investigation time. Finally, close the loop. A robust CAPA process that addresses root causes, not symptoms, prevents the same issue from consuming resources again. Efficiency is often just the visible result of a quality system that works.

Using AI to Anticipate Quality Risks

Artificial intelligence is the one I am watching most closely. Used well, it can change quality from a reactive function into a predictive one.

Consider the volume of data a modern quality system generates: batch records, deviations, laboratory results, environmental monitoring and supplier performance. Machine learning can surface patterns across those datasets that a person reviewing them one at a time would struggle to see. That opens the door to earlier detection of process drift, smarter prioritization of investigations, faster review of documentation and more targeted supplier oversight. Natural language tools can also help teams draft and summarize investigations and search decades of institutional knowledge in seconds.

There is an important caveat. In a GxP environment, AI has to be validated, transparent and governed. We need to understand what a model does, how it was trained, and how its outputs are verified, and a qualified person must remain accountable for the decision. Data integrity principles apply to AI just as they do to any other system. Regulators are actively developing expectations here, and I would rather help shape good practice than react to it.

Alongside AI, I expect continued movement toward electronic quality management systems, continuous process verification and real-time data connectivity with partners. These work best together: clean, connected data is what makes AI useful.

Leading Quality with Curiosity and Purpose

First, build deep technical credibility. You will make better decisions, and earn more trust, if you understand the science and the process behind the product.

Second, learn to be a partner, not just a gatekeeper. The best quality leaders say ‘here is how we can get to yes safely’ as often as they say no. That requires understanding the business and the pressures your colleagues face.

Third, develop your communication skills. You will often need to explain a complex risk to a scientist, an executive, and an inspector in the same week, and each needs a different framing.

Finally, stay curious and keep learning. Regulations, technologies, and expectations keep evolving, and the people who thrive are those who treat that as an opportunity. Above all, remember what the work is for. Every decision in quality ultimately comes back to the patient on the other end.