The agency’s sweeping rollout and staff challenge underscore rising momentum behind agentic AI: advanced, multiagent systems now fueling early pilots in medical writing, patient engagement and regulatory workflows across the industry.
The FDA’s recent announcement of the deployment of agentic AI capabilities for all agency employees underpins the life sciences move toward the next evolution of AI capabilities.
“Agentic AI refers to advanced artificial intelligence systems designed to achieve specific goals by planning, reasoning, and executing multi-step actions,” according to an agency press release issued December 1. “These systems incorporate built-in guidelines — including human oversight —to ensure reliable outcomes.”
As part of the agentic AI deployment, the agency has launched a two-month Agentic AI Challenge for staff to build agentic AI solutions and demonstrate them at the FDA Scientific Computing Day in January 2026.
Industry Adoption
A discussion panel at the recent Jeffries Global Healthcare conference in London also examined agentic AI capabilities. Panel member Chris Meier, managing director and partner for Boston Consulting Group, who leads the company’s AI team, said in an interview with BioSpace that agentic AI systems are more sophisticated, integrated algorithms than standard AI systems. Typically, these involve stacking of several large language models (LLMs) that can carry out particularly complex tasks, such as statistical analysis, or provide deeper insight than standard AI.
The FDA release noted the agency’s agentic AI deployment “will enable FDA staff to further advance the use of AI to assist with more complex tasks, such as meeting management, premarket reviews, review validation, postmarket surveillance, inspections and compliance and administrative functions.”
BCG developed a multiagent AI system for a pharma company to address the time-intensive task of medical writing for a single trial protocol. The tool aimed to dramatically cut time to first draft while ensuring regulatory compliance and high-quality output. According to BCG, “[t]he in-house senior medical writers have given the new approach their seal of approval, judging that no scientific rigor is lost during the process.”
A BCG-sponsored study in the journal Clinical Trials concluded off-the-shelf LLMs can reasonably generate first-draft versions of protocols, consent forms and clinical study reports. But with additional engineering of the LLMs, the writing quality can become even higher.
Daiichi Sankyo’s Agentic AI Use
In terms of additional pharma company adoption, Daiichi Sankyo is not using agentic AI or generative AI to cut employee costs, but to better answer and personalize incoming patient and healthcare practitioner queries, said Dr. Michael Zaiac, head of Medical Affairs Oncology, Europe and Canada, to BioSpace. Zaiac, who also spoke on the Jeffries panel, said Daiichi has collaborated with BCG for AI agentic integration into existing legacy systems, like Veeva, a cloud-based software for the life sciences industry.
Daiichi’s intended goal next year is to expand agentic AI use for content generation, such as internal training documents and market access dossiers, as well as medical and regulatory affairs such as protocol writing, Zaiac said.
The pharma company is using the agentic AI system in Europe and Canada and is in conversations with colleagues in the U.S., Japan and elsewhere in Asia to see if it will broaden the program’s scope there to standardize and harmonize replies, Zaiac said.
In such a heavily regulated environment, pharma and biotech are inherently cautious about wholesale adoption of agentic and even generative AI, Zaiac said. “We embrace novelties, but not just for the purpose of the novelty,” he said. Before launching the agentic AI, the company spent six weeks writing the code for the program and then had nine months of discussion with its legal team to ensure compliance. “We had to really think through what the implications of this could be,” he added.
