PHARMA LAB: Agentic AI has become a major topic across life sciences and laboratory automation. How would you define agentic AI in the context of R&D environments, and why is it generating so much attention now?
HAL WEHRENBERG: Technologies such as AlphaFold and ChatGPT have already demonstrated the potential of AI to accelerate research and improve productivity. Agentic AI takes that a step further by combining information from scientific literature, proprietary datasets, laboratory operations, and experimental results to generate insights and recommend actions.
In an R&D environment, that means moving beyond analysis toward systems that can help design experiments, guide workflows, and continuously learn from new data. The excitement comes from the possibility of dramatically accelerating discovery while allowing scientists to spend less time on repetitive tasks and more time focused on scientific innovation and decision-making.
PHARMA LAB: Laboratory automation has traditionally focused on improving efficiency and throughput. How does agentic AI represent a shift beyond these conventional automation approaches?
HAL WEHRENBERG: Traditional laboratory automation has focused on increasing throughput, reducing time from sample to result, and minimizing human error. Agentic AI expands that focus from optimizing individual steps to optimizing the entire workflow.
By connecting experimental design, laboratory execution, results, and operational metadata, AI can help identify patterns, refine future experiments, and continuously improve outcomes. The opportunity is not simply to automate more tasks, but to create a more adaptive R&D process that learns from every experiment and helps accelerate scientific progress.
PHARMA LAB: One of the most interesting discussions around AI in laboratories is the move from reactive to proactive systems. What does this mean and what does this transition look like in practice?
HAL WEHRENBERG: Historically, laboratory systems have been largely reactive, helping scientists understand what happened after a run failed or data quality was compromised. Proactive systems seek to identify conditions that may lead to problems before they occur.
This can include analysing historical performance data, environmental conditions, workflow patterns, and operational metrics to detect early warning signs. Rather than simply reporting issues, AI can help laboratories anticipate them and take corrective action before experiments are affected. The shift from reactive to proactive operations has the potential to improve efficiency, reduce wasted samples and resources, and increase confidence in experimental outcomes.
PHARMA LAB: How do you see AI helping laboratories improve robustness and reliability in workflows involving highly sensitive clinical or biological samples where even small deviations in process can have major consequences?
HAL WEHRENBERG: Biological workflows are influenced by a wide range of factors, including environmental conditions, reagent storage, instrument settings, and the timing between process steps. Modern laboratories generate large volumes of data related to these variables, but it can be challenging to understand how they interact. One of AI’s key strengths is its ability to identify meaningful patterns within complex datasets.
By connecting environmental, operational, and experimental information, AI can help uncover hidden sources of variability, improve data quality, and provide deeper insight into factors affecting assay performance. This has the potential to improve reproducibility and reduce the risk of costly failures in sensitive workflows.
PHARMA LAB: Many organisations are now exploring how AI can be integrated into existing lab infrastructure, but what challenges do laboratories face when applying AI to legacy systems and workflows?
HAL WEHRENBERG: The biggest challenge is often not the AI itself but the integration of data across disconnected systems and physical workflows. Many laboratories have instruments, software platforms, and manual processes that were never designed to work together.
The encouraging reality is that organisations do not need to completely rebuild their infrastructure to begin exploring the value of AI. Meaningful results can often be achieved by connecting existing systems, focusing on high-value use cases, and improving data accessibility. The most successful initiatives tend to start with clearly defined problems and scale from proven outcomes rather than attempting a wholesale transformation from day one.
PHARMA LAB: Looking ahead over the next three to five years, where do you expect agentic AI to have the greatest impact across pharmaceutical and biotech R&D?
HAL WEHRENBERG: The greatest impact will likely come from reducing the time required to move from scientific concept to clinical evaluation while improving the quality of candidates entering the development pipeline.
As AI accelerates hypothesis generation, candidate selection, and data interpretation, experimental validation may increasingly become the limiting factor. At the same time, scientists will remain central to defining novel experimental approaches and developing new methodologies. AI is unlikely to replace scientific creativity, but it can significantly accelerate the adoption and application of new discoveries.
Over the next five years, I expect the conversation to shift away from AI as a standalone technology and toward how connected, data-driven workflows can help researchers make better decisions and bring new therapies to patients more efficiently.
