The US Food and Drug Administration has announced major advancements in its ongoing digital modernisation initiative with the launch of Elsa 4.0, a significant upgrade to the agency’s internal artificial intelligence platform, alongside the rollout of a new consolidated data infrastructure named HALO (Harmonised AI & Lifecycle Operations for Data).

The upgraded Elsa 4.0 platform is now available across the FDA workforce, including scientific reviewers, investigators and regulatory staff. The AI tool is designed to improve operational efficiency by automating repetitive tasks and supporting data-driven decision-making across the agency.

As part of the modernisation effort, the FDA has consolidated more than 40 separate application systems, data portals and submission platforms used across its various centres into HALO. The unified platform is expected to strengthen the deployment of AI-powered workflows by enabling staff to directly access and analyse regulatory data without manually uploading documents into individual systems.

FDA Commissioner Marty Makary said the latest upgrades reinforce the agency’s position as a leader in the adoption of AI technologies within healthcare regulation. He noted that reducing administrative burdens would allow scientific teams to focus more on research and evaluation while improving workflow efficiency.

Jeremy Walsh, Chief AI Officer at the FDA, stated that the integration of Elsa with HALO marks a major shift in how the agency interacts with data. According to Walsh, Elsa will now function as a central access point to FDA systems and datasets, enabling more advanced AI-driven regulatory science and potentially accelerating the delivery of treatments and therapies to patients.

Since the launch of Elsa 1.0 in June 2025, the FDA’s Office of Digital Transformation has continued expanding the platform’s capabilities and testing new applications in collaboration with agency reviewers.

The newly launched Elsa 4.0 introduces several enhanced features, including custom AI agents, automated document generation, quantitative data analysis and chart creation, secure web-search capabilities, voice-to-text dictation, optical character recognition for scanned files, improved chat functionality and advanced search tools for navigating large document repositories.

The FDA stated that Elsa operates within a FedRAMP High secure Google Cloud Platform environment and does not train on confidential industry-submitted data, ensuring protection of sensitive research and regulatory information. Although the system can access refreshed secure web data, it is not directly connected to the public internet. The agency also emphasised that FDA personnel remain actively involved throughout the AI workflow process to verify inputs, analytical methods and implementation outcomes.

The initiative forms part of the FDA’s broader efforts to modernise regulatory operations and improve the efficiency of evaluating drugs, vaccines, medical devices and other healthcare products.