Scaling Organoids for Drug Discovery: How Automation and AI Are Transforming the DMTA Cycle

The rapid acceleration of compound generation through artificial intelligence is reshaping priorities across the drug discovery landscape. As AI and automation continue to increase the speed at which new molecules can be produced, the primary bottleneck is shifting from molecule generation to the quality of biological models used to evaluate those compounds. Organoids—three-dimensional, multicellular tissue models that recapitulate key aspects of human organ biology—are emerging as one of the most promising solutions to this challenge. However, experts caution that for organoids to fulfill their potential, they must evolve from bespoke laboratory models into scalable, standardized platforms suitable for industrial drug discovery.

Bridging the translational gap in the DMTA cycle

Organoids are steadily establishing themselves as a valuable testing methodology within the design-make-test-analyze (DMTA) cycle that underpins modern drug discovery. By occupying the space between the relative simplicity of spheroids and the complexity and cost of in vivo models, organoids offer an attractive balance. Their ability to generate data more representative of human biology than conventional two-dimensional cell cultures makes them particularly valuable during lead optimization, where efficacy must be carefully balanced against potential toxicity.

“3D models such as organoids are increasingly important because they offer greater human relevance than traditional 2D cell culture systems,” said Vicky Marsh Durban, PhD, Director of Human Relevant Models at Molecular Devices (搜索)搜索该词条. “This improved biological relevance enables better translatability of findings into clinical settings and ultimately into patients.”

Liver organoids, for example, have the potential to identify hepatotoxicity much earlier in the discovery process, before substantial time and resources have been invested in a lead series. Similarly, tumor organoids—particularly those derived directly from patients—can provide more predictive insights into therapeutic response in oncology. Intestinal organoids may also offer a more human-relevant platform for assessing drug absorption, barrier integrity, and gastrointestinal toxicity than traditional two-dimensional assays.

The scalability bottleneck

Despite their promise, organoids face a fundamental challenge: scalability. Currently, organoids are generally produced in relatively small batches by highly skilled biologists, with outcomes often influenced by individual expertise and laboratory-specific practices. While this approach is appropriate for exploratory research, it is inherently difficult to scale. Future DMTA workflows may require thousands of highly comparable organoids to be generated on demand—a requirement that traditional artisanal manufacturing methods were never designed to achieve.

“What's important to remember is that each source presents unique scaling challenges,” Marsh Durban noted. For iPSC-derived organoids, differentiation can take many weeks or even several months, during which cells require multiple media changes, tightly controlled culture conditions, and careful handling. Patient-derived organoids introduce additional difficulties, as these models are often cultured within hydrogel matrices that can be highly viscous and more difficult to handle than standard liquid cultures.

Greg Walkowiak, Consultant and Project Leader at TTP, draws a parallel to the biologics industry. “Just as biologics transitioned from small-scale laboratory production to tightly controlled industrial manufacturing, organoids may require a similar transformation if they are to become dependable tools within mainstream drug discovery,” he explained.

Automation as the path to reproducibility

Automating organoid manufacturing has the potential to address both scalability and consistency simultaneously. Automated systems standardize the many manual steps involved in cell culture—media changes, cell handling, and other routine manipulations—that traditionally create opportunities for variability.

“Most laboratories have experienced situations where a particular scientist develops exceptional expertise with a specific model system,” said Marsh Durban. “Automation coupled with AI-driven software—such as the CellXpress.ai Automated Cell Culture System—enables organizations to capture that expertise and apply it consistently across every culture.”

To support automated organoid manufacturing, microfluidic technologies could be combined with robotics and other automated liquid-handling systems to reduce both labor requirements and operator variability. AI-enabled imaging techniques and other automated functional assays could further reduce reliance on human judgment during both in-process monitoring and final quality control assessments.

Beyond improving efficiency, perhaps the greatest advantage of automation is that it shifts organoid manufacturing away from a process dependent upon individual expertise toward one governed by repeatable manufacturing controls. In the long term, this increase in reproducibility may prove even more valuable than the labor savings achieved through automation.

AI-driven quality control and monitoring

AI can play a major role in improving culture quality when integrated with automated imaging systems. By continuously capturing images throughout the culture process, researchers can train machine learning models to identify abnormalities and inconsistencies that may indicate emerging problems. These systems use large historical datasets to recognize patterns that might otherwise go unnoticed, flagging potential issues much earlier and reducing the risk of failed experiments later in the workflow.

Routine monitoring and comprehensive data capture are essential components of any automated culture platform. “Every image, event, and process step contributes to a historical record that can be used to investigate unexpected outcomes,” Marsh Durban explained. “If a culture fails or produces results that differ from expectations, researchers can trace the entire process back to the beginning and identify what changed.”

Assay-ready organoids and storage challenges

Even with industrial-scale manufacturing, storage remains a critical enabling technology. Organoids typically require weeks or even months to mature, while also having a relatively limited period during which they remain viable for experimentation. The most widely used method for biological storage today is controlled-rate freezing, which employs cryoprotective agents such as DMSO to minimize damage caused by ice crystal formation. However, preserving mature organoids presents a greater challenge than preserving individual cells because organoids are large, structurally complex three-dimensional tissues.

An alternative approach, known as vitrification, employs higher concentrations of cryoprotectants together with ultra-rapid freezing to eliminate ice crystal formation entirely. Although promising, additional research is needed to overcome challenges associated with osmotic stress and the potential toxicity of cryoprotective agents.

Assay-ready organoids represent one practical solution already being deployed. These provide researchers with standardized batches of organoids that can be used directly from frozen storage. “Because an entire batch is generated simultaneously using the same operators, protocols, and reagents, variability is minimized from the outset,” said Marsh Durban. “All organoids within a batch are at the same passage number and have been cultured under identical conditions.”

Enabling earlier decision-making

The ability to generate complex 3D models at scale enables high-throughput screening using biologically relevant systems much earlier in the drug discovery process. This allows researchers to evaluate efficacy and toxicity using more predictive models before significant resources have been invested. If a candidate demonstrates poor efficacy or unexpected toxicity, those issues can be identified sooner, allowing teams to discontinue unsuitable programs earlier.

“A candidate compound that performs well in a conventional cell-based assay but shows early signs of liver toxicity or poor efficacy in a human-relevant organoid model could be deprioritized before proceeding to expensive animal studies and advanced development programs,” Walkowiak noted. This aligns closely with the industry's “fail fast, fail cheap” philosophy.

Following the passage of the FDA Modernization Act 2.0 in 2022, regulatory agencies and funding organizations have increasingly supported organoids as part of the broader family of new approach methodologies (NAMs) designed to reduce reliance on animal testing throughout drug discovery and development.

The future: manufacturing as the key determinant

The long-term success of organoids will depend not solely on advances in biology, but equally on advances in manufacturing. Once organoids can be produced, qualified, stored, and distributed with the same reliability as other essential laboratory consumables, they will transition from specialized research tools into core infrastructure supporting modern drug discovery.

“In many situations, consistency may prove to be more valuable than achieving the highest possible level of biological complexity,” Walkowiak emphasized. As artificial intelligence becomes increasingly embedded within discovery workflows, the ability to generate large, consistent datasets may ultimately prove just as valuable as the biological models themselves.

Together, automation and AI form a complementary cycle: automation generates reliable, standardized data, while AI extracts deeper insights from those datasets. As both technologies mature, they are likely to play a central role in improving the efficiency, scalability, and predictive power of future drug discovery programs.