AI-assisted design is playing an increasingly important role in how biologic drug candidates are developed, and companies like AstraZeneca are actively building their engineering teams to further this. “Everything we do – design, manufacturing, testing and analysis – is now computer-enhanced,” said Puja Sapra, AstraZeneca’s senior vice president and head of research, development, biologics engineering and oncology target discovery. “Cycle times are decreasing while productivity and innovation are increasing.”
Sapra explains that AstraZeneca’s approach follows a build-measure-learn loop. AI computationally generates or prioritizes candidate molecules and predicts which designs are most likely to be successful. Scientists then focus their lab resources only on the top candidates. This creates tighter feedback cycles, fewer dead ends, faster iteration, and allows medicine to pursue disease targets previously considered untreatable. The number of possible molecular combinations far exceeds what human teams can systematically explore, so using AI to narrow down and refine test options has become a major focus of biologic drug design.
Solving complex drug design problems
Beyond timeline acceleration, AI is also being applied to the discovery of entirely new classes of medicines. Traditional biologics typically target one disease pathway. Next-generation drugs can attack multiple targets simultaneously or precisely deliver therapeutic payloads to specific cells. To achieve this, many variables need to be optimized at once. Looking to the future, Puja Sapra explains that AI-driven models can help design increasingly complex, multi-specific biologics. “For example, such models can help identify which two or three targets to prioritize based on the underlying biology and optimize across multiple parameters to balance efficacy, stability, manufacturability, and safety of the molecule.” “Using drugs for things that cannot be drugged is becoming a reality,” Sapra says. “These technologies will ultimately enable the development of medicines to achieve goals previously thought unattainable. The potential for patient benefit is remarkable.”
data moat
McKinsey estimates that generative AI, combined with other computational tools, has the potential to shorten drug discovery timelines by up to 50%. But every AI model is only as good as its training data. In drug discovery, this means large amounts of high-quality biological data. Experiments provide a rich source of such data. Whether successful or unsuccessful, each experiment generates signals about what works and what doesn’t.
“Data is our differentiator,” Sapra says, explaining how the company’s dataset is unique and multimodal and includes molecular structures, binding measurements, safety profiles, and manufacturing results. “We have built an intentionally diverse portfolio across multiple disease areas and drug types. All that data allows us to fine-tune our Frontier AI models with richer and more representative training sets,” she continued.
Building an autonomous detection engine
To bring all this data together in one place, AstraZeneca is building a so-called “Lab of the Future” facility in Kendall Square in Cambridge, Massachusetts, where AI and robotic automation can form a continuous closed-loop detection system. “While self-driving cars use sensors and models to navigate their environments, this system uses AI to make predictions, robotic systems to run experiments, and equipment to generate data,” Sapra explains. That data is fed directly back into the model to accelerate each subsequent cycle.
“Throughout, scientists remain central to the process, providing oversight, judgment, and strategic direction that ensures outcomes are explainable, acceptable, and directed toward potential patient benefit,” she added.
Eventually, automated high-throughput systems will be able to generate and evaluate thousands of molecular interactions every week. “This generates AI-enabled data at a scale that traditional workflows cannot accommodate,” Sapra says. “Robotic sample processing, automated quality checks, and integrated data pipelines can also help significantly accelerate early drug development timelines.”
