Velocia: AstraZeneca's Agentic AI for Medical Evidence Generation

Written by:

Medical evidence generation is an essential part of developing a new medicine. The integration of AI tools in clinical research is helping drive efficiencies; however, fragmentation remains a key issue. Velocia is a major new AI initiative that brings AstraZeneca’s evidence expertise, data and purpose-built agentic AI solutions into one connected way of working. Velocia is designed to enhance the evidence generation process through higher quality, greater capacity and faster paths from study design concept to practice-changing impact for patients.

Why is medical evidence generation so complex?

Evidence generation is one of the most complex parts of the drug development process. A research question may move through planning, delivery, data cleansing, analysis, reporting and publication before it can inform practice. Every step relies on critical decisions made earlier in the process, with each stage often happening in siloes within specialist teams.

Specialisation is essential. Fragmentation is not. When teams, tools and data are separate, context can be lost during handovers. People might reconstruct earlier decisions, replicate completed work or repeat the same mistakes. A delay at one point can travel through the rest of the study, while learnings can be missed and not fed back into the chain.

How is Velocia streamlining the evidence generation process?

Velocia is AstraZeneca’s solution to these challenges. It is a major new AI initiative built as a connected ecosystem of multi-agent solutions — bespoke AI tools that are individually designed to handle complex and specific tasks within the evidence generation journey. Together, they create a flow of knowledge in which the output from one stage can inform others, providing live, continuous and predictive intelligence across the evidence generation lifecycle.

The ambition goes beyond adding another narrowly focused AI agent or digital tool. With human oversight throughout, Velocia is designed to support higher quality, greater capacity and a faster, more effective path from study concept to evidence that can change medical practice.

Definition: Agentic AI. Agentic AI refers to systems made up of agents that can plan and carry out specific tasks towards a goal with a degree of autonomy. In Velocia, these agents work as connected multi-agent solutions, with human oversight throughout.

What differentiates Velocia’s AI approach to evidence generation?

Velocia is purpose-built for AstraZeneca’s evidence generation processes, with the company’s data, standards and governance embedded from the outset.

Velocia is built on AstraZeneca’s enterprise AI and data platforms. This provides a foundation of AstraZeneca’s extensive clinical and real-world data, including information from contract research organisations. The use of real-world data means studies are designed to better reflect the populations being targeted with a treatment or intervention. Different language models can be selected for different tasks as the technology evolves, shifting from isolated AI assistance to connected AI capability.

In 2026, AstraZeneca is validating Velocia’s first group of interconnected multi-agent solutions on real studies, with further solutions planned for development and launch through 2027 and beyond. Velocia is expanding at scale, while adapting alongside advances in technology and implementing improvements based on direct learnings.

Built on data, governed by people

With Velocia, people remain in charge, with each step overseen by people who have years of experience. This partnership keeps compliance and specialist judgement with AstraZeneca’s experts, while AI extends the capacity to act on them faster.

That distinction is important because the use of agentic AI in clinical research and healthcare is still relatively new. A 2026 academic review of agentic AI in healthcare found that most studies evaluating the tools lacked robust clinical validation, and in their assessment only one trial involved patients.¹

AstraZeneca’s risk-based AI Governance Framework underpins Velocia, with human oversight, traceability and review built into its development and use. This follows the human-centric, risk-based approach outlined in the joint European Medicines Agency (EMA) and US Food and Drug Administration (FDA) Guiding Principles of good AI practice in drug development, published in January 2026.²

Measuring the value of AI in evidence generation

The success of Velocia does not come down to the number of AI agents or single examples of improvements it can produce. The meaningful tests are whether evidence generated can become more useful and representative, whether issues are identified earlier, whether studies can be delivered with greater quality and pace, and whether time (and therefore cost) released from repetitive work is successfully redirected.

Simply put, the additional capacity created allows teams to tackle more complex questions and focus on the work that requires human expertise. The cost and time savings have value when they are converted into more and better evidence, not when they are treated as ends in themselves.

The change is organisational as well as technical. Global workforce research points to significant demand for reskilling teams as AI changes how work is designed, while continuing to emphasise human capabilities.³ For Velocia, that means giving teams the confidence and skills to work with new tools, and the judgement to question them.

Leadership in this field will come from organisations that connect technology to real work, build trust into every step and prove value through evidence.

Velocia is AstraZeneca’s approach to doing exactly that: turning a fragmented journey into a connected, human-led evidence ecosystem, and moving with purpose from data to decisions and, ultimately, into impact for patients.

Veeva ID: Z4-89635
Date of preparation: October 2026