FDA Deploys Agentic AI to Overhaul National Safety Monitoring System

HERNDON, VA – April 16, 2026 – The U.S. Food and Drug Administration (FDA) has officially launched a landmark modernization of its public health surveillance infrastructure, powered by an advanced artificial intelligence platform from ThinkTrends. The new nationwide Adverse Event Monitoring System (AEMS) went live this week, replacing a patchwork of aging databases with a single, real-time system designed to protect American consumers by more effectively monitoring the safety of drugs, vaccines, medical devices, foods, and cosmetics.

This deployment marks a pivotal moment for the federal agency, which processes an estimated six to seven million adverse event reports annually. The new AI-driven system is at the heart of an initiative projected to save taxpayers approximately $120 million over the next five years while dramatically increasing the speed and accuracy with which regulators can detect and respond to potential safety threats.

A New Era for Public Health Surveillance

The launch of AEMS addresses long-standing criticisms of the FDA's previous safety monitoring capabilities. For years, the agency relied on multiple, siloed legacy systems—such as the FDA Adverse Event Reporting System (FAERS) and the Vaccine Adverse Event Reporting System (VAERS)—that were notoriously difficult to use and analyze collectively. These fragmented systems cost the agency roughly $37 million annually to maintain.

In a recent statement, FDA Commissioner Marty Makary, M.D., M.P.H., described the previous infrastructure as "outdated and fragmented," noting that it wasted taxpayer dollars and created dangerous "blind spots" in post-market surveillance. The new AEMS consolidates these disparate data streams, along with reports from systems monitoring medical devices (MAUDE), human foods (HFCS), and tobacco products (CTPAE), into one unified environment.

"The team executed with perfection and delivered the biggest technical transformation in agency history," said FDA Chief AI Officer Jeremy Walsh, calling the achievement "the new FDA." The system is designed not only to serve agency scientists but also to provide greater transparency to researchers and the public through a highly interactive web-based dashboard.

The 'Agentic AI' Engine

At the core of this transformation is ThinkTrends' agentic Document AI platform. Unlike basic automation tools, 'agentic AI' refers to sophisticated systems designed to understand goals, reason through multi-step plans, and execute complex tasks with built-in safeguards and human oversight. In this application, the AI acts as an intelligent agent for processing vast amounts of unstructured safety data.

The FDA receives adverse event reports in a wide variety of formats, from emailed forms and scanned PDFs to faxes and even handwritten documents. Previously, manually processing this data was a slow, labor-intensive, and error-prone endeavor. ThinkTrends' platform automates this entire intake process. The AI intelligently ingests the reports, classifies them, extracts critical information, and converts it all into a standardized, structured format that can be instantly analyzed.

This advanced capability is crucial for accelerating the agency's ability to identify emerging safety patterns. By automatically coding and normalizing millions of reports, the AEMS can flag potential issues in near real-time, enabling a shift from reactive damage control to proactive public health protection.

"Modernizing adverse event monitoring at a national scale required consolidating multiple legacy systems, handling millions of safety reports, and delivering a reliable platform on an accelerated timeline," said Jyotiska Biswas, CEO of ThinkTrends. “We are proud to support the U.S. Federal Health Agency in deploying a system that strengthens real-time safety surveillance for regulators, researchers, and the public.”

The High Stakes of Regulated AI

Deploying a sophisticated AI within one of the nation's most critical regulatory bodies is a task fraught with immense technical and compliance challenges. The system must handle sensitive and protected health information (PHI) with uncompromising security while adhering to a stringent web of federal regulations, including HIPAA, the Federal Information Security Management Act (FISMA), and the guidelines set forth in the NIST AI Risk Management Framework.

ThinkTrends, which specializes in AI for regulated sectors, designed its platform for this high-stakes environment. The system operates within a secure AWS GovCloud architecture aligned with FedRAMP standards. To ensure data privacy and system integrity, the platform incorporates enterprise-grade "guardrails" that automatically redact or block personally identifiable information (PII) and PHI. The AI models do not train on any input data, protecting the sensitive commercial and patient information submitted by the industry.

Furthermore, the system is built around a principle of human-guided oversight. The AI assists FDA personnel by automating rote tasks and providing powerful analytical tools, but final decisions and actions remain in the hands of human experts. This combination of intelligent automation and expert oversight serves as a blueprint for how AI can be responsibly integrated into other regulated industries, from finance to life sciences.

Measuring Success: From Cost Savings to User Engagement

The impact of the AEMS is being measured not only in future cost savings but also in immediate operational improvements. The projected $120 million in savings over five years represents a significant return on investment compared to the costly upkeep of the previous systems. But perhaps more telling is the data from the platform's early pilot phase.

During pilot testing, the new system saw a staggering 3,000 percent increase in user activity compared to the legacy reporting methods. This dramatic spike suggests that the unified, intuitive interface has removed significant barriers to reporting and data access, encouraging more robust engagement from both internal and external stakeholders. By making the system easier to use, the FDA is likely to capture more data, leading to a richer and more complete picture of product safety.

The broader goal is to foster a more data-driven and transparent regulatory process. By centralizing adverse event data, consumer complaints, and whistleblower submissions in one place, the AEMS provides the FDA with an unprecedented ability to connect dots and identify safety trends across its entire regulatory portfolio. This successful deployment is poised to influence how government agencies approach digital transformation, demonstrating that strategic investments in trustworthy AI can yield substantial benefits for both fiscal responsibility and public welfare.