From AI to Engineered Biology: What’s Next for Drug Discovery?
Advances in technology point to a future where drug discovery is faster and more human-relevant.
Technologies such as high-throughput screening, multiomics, and computational modeling have helped transform drug discovery over the last two decades. Now, AI, engineered biology, and increasingly predictive human datasets are beginning to reshape not only what scientists can design, but also how discovery programs are organized and advanced.
As Head of Pipeline and Commercialization Strategy at Constructive Bio, Dr. Saleha Patel has witnessed this evolution firsthand. Patel leads efforts to develop new-to-nature peptide and protein therapeutics by utilizing full genomes, engineered translation, and scalable biomanufacturing.
Ahead of chairing the keynote discussion at ELRIG Drug Discovery 2026 on “20 Years of Discovery: Lessons from the Past, Blueprints for the Next 20,” Patel discussed four major shifts in modern drug discovery, the growing convergence of computational and experimental science, and the technologies that could define the field’s next two decades.
Four shifts shaping the future of drug discovery
The adoption of AI is one of the most visible changes in drug discovery. Generative models can propose potential compounds or biological designs, while active learning systems can use experimental results to determine which tests would be most informative next.
Patel identified AI-enabled target and compound discovery as the first of four dominant shifts: “Four shifts dominate. One: the rise of AI-enabled target and compound discovery, where generative models and active learning are compressing design–make–test–analyze cycles and reducing the number of synthesis rounds needed.”
The second shift is a move away from selecting a therapeutic format before fully understanding the underlying biology. Instead, modality-agnostic strategies allow teams to select an approach best suited to the target and clinical context.
Modality-agnostic drug discovery
Modality-agnostic drug discovery is an approach in which researchers evaluate multiple therapeutic formats and select the modality that best matches the biological mechanism, target properties, and intended clinical application.
The third shift is the earlier integration of human evidence. Large cohorts, real-world data, and multiomics can help researchers prioritize targets with stronger links to disease biology. This is intended to reduce the risk of advancing candidates that perform well in experimental models but prove less relevant in patients.
Finally, Patel highlighted the rise of platform companies and venture-creation models that build assets around core technologies, rather than chasing single programs. “This has changed how capital is deployed and how risk is managed across portfolios,” she said.
Uncovering Hidden Binding Sites in Challenging Drug Targets
Explore how accelerated long-timescale molecular dynamics can uncover metastable protein conformations and cryptic binding sites missed by static structures, with an Eotaxin-1 case study demonstrating their application in small molecule discovery.View Webinar
Four shifts shaping how modern drug discovery is performed:
- AI is accelerating iterative molecular and target discovery.
- Modality selection is increasingly guided by biology and clinical need.
- Integration of multiomics and real-world data is moving earlier in the discovery process.
- Platform strategies are changing how companies construct portfolios and manage risk.
Closed-loop discovery connects computation and experimentation
Traditional linear drug discovery moves slowly from computer models to lab tests. A model generates a prediction, experimental scientists test it, and the resulting data are reviewed before another computational cycle begins.
That boundary is becoming less distinct. Patel described an emerging model in which computation and experimentation operate as a continuous feedback loop: “We’re moving from a linear ‘compute then test’ model to tightly coupled, iterative loops where computation guides experiment and experiment retrains computation.”
Ask The Expert: Navigating the Path From Discovery to Clinical Impact
Join Dr. Stella Vnook (30+ years biotech/pharma) as she explores navigating discovery to clinical impact - overcoming barriers, mitigating risk and advancing therapies to patients.View Ask The Expert
In a closed-loop workflow, models prioritize experiments based on their expected information value. The resulting data are then fed back into the model, allowing it to refine subsequent predictions. According to Patel, active learning and closed-loop platforms are already producing “30–50% timeline reductions in specific applications by focusing resources on the most informative experiments.”
Active learning
Active learning is an iterative feedback process in which a model identifies valuable data based on model-generated assumptions and uses this data to enhance its performance.
Speed, however, is only one part of the change. These systems depend on organizational integration, shared data standards, and reliable experimental records. Computational and laboratory teams must work closely enough to ensure that model outputs are experimentally actionable and that results can be returned in a consistent, machine-readable form.
Negative data also become more valuable in this environment. A failed experiment can establish boundaries, expose an incorrect assumption, and help a model avoid unproductive regions of chemical or biological space.
The broader objective is not simply to run the existing discovery process faster. “For modern drug discovery, it means faster hypothesis testing, better prioritization of targets and chemistries, and ultimately a higher probability of identifying candidates with a clear mechanism and clinical rationale,” said Patel.
Tech Spotlight: AI and CRISPR Genome Editing 2026
Explore how AI, protein engineering and high-throughput screening are advancing CRISPR design, with insights into engineered CasX, gRNA optimization, genome editing, epigenetic silencing and translating computational designs into practical workflowsView Webinar
Requirements for effective closed-loop drug discovery:
- Computational and experimental teams must operate as an integrated unit.
- Standardized, high-quality data are essential for model retraining.
- Negative results must be captured and treated as informative evidence.
- Experiments should be selected for the knowledge they can generate, not only their likelihood of producing a positive result.
Engineered biology and AI could expand therapeutic design
Over the next 20 years, Patel expects drug discovery to become more programmable. Advances in engineered biology could enable researchers to design therapeutic functions that are difficult or impossible to achieve using naturally occurring biological components.
Programmable cells, synthetic immune circuits, and in vivo gene editing are among the approaches that could contribute to this shift. Genetic code expansion and engineered translation systems may also expand the chemistry available to peptide and protein therapeutics, allowing researchers to tune properties such as stability, specificity, and biological activity.
Accelerate Your PROTACs Drug Discovery and Development
This eBook shows how an integrated, end-to-end analytical approach helps labs generate reliable, regulatory-ready PROTAC data at every stage of development.View eBook
AI-driven discovery is expected to develop in parallel. Its role could extend beyond predicting molecular structures to coordinating fuller design–make–test–analyze cycles for molecules, proteins, and delivery technologies. Greater model interpretability and better-quality training data will be critical if these systems are to become trusted elements of routine scientific decision-making.
“As data quality improves and models become more interpretable, we’ll see AI as a standard co-pilot in discovery teams rather than a niche capability.” — Dr. Saleha Patel.
Patel also highlighted the growing role of predictive science anchored in human data. Large and diverse cohorts linked to multiomics and longitudinal health records could support more precise biomarkers and patient-stratification strategies.
“That’s how we move from ‘does it work on average?’ to ‘for whom, and under what conditions, does it work best?’, which is the foundation of precision medicine at scale,” explained Patel.
“Together, these trends point to a future where discovery is faster, more human-relevant, and more tightly connected to the realities of clinical care and public health,” she added.
Tech Spotlight: High-Throughput Screening 2026
Explore reliable high-throughput screening workflows for vaccine and antibody discovery, covering assay design, validation and data analysis to improve reproducibility, data quality and candidate prioritization.View Webinar
Technologies shaping the future of drug discovery:
- Engineered biology could create therapeutic functions and chemical diversity not found in nature.
- Automated design–make–test–analyze systems could make AI a routine discovery tool.
- Human-linked multiomics and longitudinal data could improve biomarkers and patient stratification.
Learning from drug discovery’s past to design its future
The next era of drug discovery will not be defined by one technology alone. Its success will depend on the connections between computation and experimentation, therapeutic design and manufacturing, and biological innovation and human evidence.
These intersections will form the focus of the ELRIG Drug Discovery 2026 keynote panel chaired by Patel. Bringing together perspectives from venture creation, AI-enabled biology, drug development, and engineered biology, the session will examine which lessons from the past 20 years should inform the next generation of transformative medicines.
Drug discovery is moving toward integrated platforms that combine stronger human evidence, broader therapeutic capabilities, and repeated computational and experimental learning.
Key takeaways:
- AI and active learning are changing both the speed and structure of discovery workflows.
- Engineered biology could expand the chemical and functional possibilities of peptide and protein therapeutics.
- Predictive human data may help the field progress from average treatment effects toward precision medicine at scale.
Add Technology Networks as a preferred Google source to see more of our trusted coverage.
