AI-assisted design is playing an increasingly important role in the development of biologic drug candidates, and companies like AstraZeneca are actively building their development teams to take this further. “Everything we do, whether design, manufacturing, testing or analysis, is now computationally enhanced,” says Puja Sapra, senior vice president and head of R&D for biologics engineering and targeted oncology research at AstraZeneca. “The throughput times will be shorter, productivity and innovation will increase.”
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. The scientists then concentrate their laboratory resources only on the top candidates. This results in a tighter feedback cycle with fewer dead ends, faster iteration, and the ability to pursue disease targets previously considered untreatable by medicine. Because the number of possible molecular combinations far exceeds what a human team can systematically explore, the use of AI to narrow and refine testing options has become a major focus in biologic drug development.
Table of Contents
Tackling complex drug design problems
Beyond accelerating timelines, AI is also being used to discover entirely new classes of drugs. Traditional biologics typically target one disease pathway. The next generation of drugs can hit multiple targets simultaneously or deliver therapeutic agents precisely to specific cells. Achieving this requires optimization across many variables simultaneously. Looking ahead, AI-driven models could help develop these increasingly complex, multispecific biologics, explains Puja Sapra. “For example,” she continues, “such models could help determine which two or three targets to prioritize based on the underlying biology and then optimize across multiple parameters to balance a molecule’s potency, stability, manufacturability, and safety.” “Drug addiction is becoming a reality,” says Sapra. “These technologies will ultimately allow us to develop drugs against targets we thought were out of reach. The potential for benefit to patients is remarkable.”
The data ditch
McKinsey estimates that generative AI combined with other computing tools could shorten drug development timelines by up to 50%. But every AI model is only as good as its training data. In drug research, this means large amounts of high-quality biological data. Experiments can provide a rich source of such data. Regardless of whether they are successful or not, each experiment produces a signal about what works and what doesn’t.
“Data is our differentiator,” says Sapra, explaining that the company’s data sets are proprietary and multimodal and include molecular structures, binding measurements, safety profiles and manufacturing results. “We have intentionally built a diverse portfolio across multiple disease areas and drug types. All of this data allows us to refine groundbreaking AI models with larger, more representative training sets.” She continues: “In addition, we have invested in deep screening technologies to generate additional data sets needed in large quantities to continually refine and validate our models.”
Building an autonomous recognition machine
To bring all this data together in one place, AstraZeneca is building a facility in Kendall Square, Cambridge, Massachusetts, that it calls a “lab of the future” where AI and robotic automation can form a continuous, closed-loop discovery system. “While a self-driving car uses sensors and models to navigate its environment, this system uses AI to make predictions, robotic systems to conduct experiments, and instruments to generate data,” explains Sapra. This data flows directly back into the models and accelerates each subsequent cycle.
“Scientists remain at the center of the process, providing oversight, judgment and strategic direction to ensure results are explainable, tolerable and aligned with potential patient benefit,” she adds.
At some point, automated high-throughput systems will be able to perform and evaluate thousands of molecular interactions weekly. “This generates AI-enabled data at a scale that traditional workflows cannot achieve,” says Sapra. “Robotic sample handling, automated quality checks and integrated data pipelines also have the potential to help significantly shorten early drug development timelines.”
https://www.technologyreview.com/2026/07/23/1140346/how-ai-helps-scientists-design-the-next-generation-of-medicines/
