When The Regulator Moves Faster Than The Industry It Regulates

Nagesh Nama, CEO, xLM. Nagesh Nama is a seasoned technology executive with over 30 years of experience in life sciences.

getty

​Something remarkable happened at a government press conference on April 28, 2026. The FDA—an agency more commonly associated with slow-moving bureaucracy—announced that it had conducted the first real-time clinical trial. This is an actual functioning program, with two of the world's largest pharmaceutical companies already streaming live clinical data directly into FDA review systems through a cloud platform powered by AI.

Let that land for a moment. The federal regulator has outrun the industry it oversees.

The FDA's announcement described a direct, AI-enabled, cloud-based data feed from active drug trials, meaning FDA scientists can now observe patient safety signals in near-real time rather than waiting months or years for data to move through the traditional chain. AstraZeneca and Amgen are already on board.

This is a 60-year paradigm being retired.

But the real story—the one that should keep every life sciences executive up at night—isn't the technology itself. It's the speed at which a government agency achieved transformation that the private sector has repeatedly failed to replicate. And the number that tells that story is this: According to FDA CAIO Jeremy Walsh, "About 1% of the agency’s workforce regularly used generative AI in their jobs." By April 2026, that number had crossed 80%. The agency accomplished this with a reduced workforce, no additional budget and ahead of its own deadlines.​

The tool at the center of this story is Elsa, the FDA's LLM-powered generative AI assistant. Elsa allows FDA staff to read, write, summarize regulatory documents, analyze adverse events and generate code—all within a secure, validated environment. With AI tools like Elsa, tasks that previously consumed 10 days of expert time are now completed in 20 minutes.​

The 'Elsa' Rule For The Pharma Industry

This brings us to the harder conversation. The pharmaceutical industry hasn't been sitting idle during this period. Industry surveys find that 85% of pharma executives report that adopting AI tools is a top priority. What reports also show is that 95% of corporate AI projects yield no return and that a mere 5% of pilots transition into production with quantifiable value. The industry is extraordinarily good at running controlled experiments with AI. It's far less effective at making AI part of how the organization really works on a daily basis.​​

The FDA didn't run a pilot of Elsa for two years. It deployed it agency-wide, measured adoption and held leadership accountable for the results.

The FDA Is Paving The Way For The Industry​

This matters especially in the context of manufacturing. The FDA and EMA jointly published the "Guiding Principles of Good AI Practice in Drug Development" in January 2026, establishing a foundational framework for how AI must be designed, validated and governed across the pharmaceutical product life cycle, from nonclinical research through clinical development, manufacturing and post-marketing surveillance. In my recent Forbes Technology Council article covering those principles in depth, I outlined why they represent not merely a governance checklist but a strategic inflection point for every organization in the life sciences sector. The principles demand human-centric design, risk-based validation, multidisciplinary expertise, data governance documented in a traceable and verifiable manner, and life cycle management that includes ongoing model monitoring for data drift.

Read together with the real-time clinical trial initiative, these principles send a unified signal: The regulatory environment is being rebuilt around continuous, AI-enabled, data-native operations. The FDA isn't asking the industry to do something the regulator itself is unwilling to do. It's doing it first.

​The Practical Implications For Pharma Manufacturers

For pharma manufacturers, the practical implications are direct. The current model of compliance is largely retrospective—organizations assemble batch records, deviation histories and documentation trails to prove after the fact that processes were followed. The FDA is departing from this model in its own operations, and the trajectory of its AI strategy points toward an oversight environment that is always-on, signal-based and real-time rather than episodic and document-centric. The AI in pharmaceutical manufacturing market reflects the urgency of the shift: It's projected to grow from $1.2 billion today to $34.7 billion by 2040.

​​How We Can Follow FDA's Lead

To follow in the FDA's footsteps, I suggest the following:

• Deploy one high-volume, companywide GxP workflow within 12 months, the way the FDA deployed Elsa, and measure adoption rather than novelty.

• Make QA the AI governance authority before HR or IT claims the territory.

• Rebuild the data layer before buying more agents, because agentic systems are only as compliant as the data they act on.

• Validate AI the way you validate equipment: risk-based, life-cycle-managed, monitored for drift and aligned to the FDA and EMA Guiding Principles.

• Tie executive compensation to AI adoption metrics, because 95% of corporate AI projects stall precisely where no one owns the number.

• Design for governed autonomy, not full autonomy. The advantage isn't removing humans—it's putting them in the right place.​

Conclusion​

Commissioner Makary said something at the April press conference that deserves to be quoted in every pharmaceutical manufacturing boardroom in the country: "For 60 years, we've been conducting clinical trials in the same way, where key data signals can take years to reach the FDA." He wasn't describing a technology problem. He was describing a culture of accepting what's always been done as what must always be done.

You need to ask yourself whether your organization's pace of change is fast enough to stay aligned with the regulatory environment that's already being constructed around you—and whether your workforce's daily relationship with AI tools reflects the same urgency that a government agency just demonstrated is possible.

We've seen 80% adoption in 12 months, in a federal agency with a reduced headcount and no additional budget. The private sector has fewer excuses than it's ever had.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?