The CureRX Platform
Closed-loop small-molecule discovery from target definition to synthesis-ready lead series.
CureRX runs a closed-loop discovery cycle combining generative chemistry, structure-based design, and reinforcement-learning optimization.
Step 1
Target Definition & Druggability Assessment
Biology-grounded target definition and druggability assessment
Step 2
Generative Chemistry
De novo molecular generation via structure-based design.
Step 3
Multi-Objective Optimisation
Simultaneous optimization of potency, selectivity, safety, and ADMET.
Step 4
Lead Delivery
Experiment-ready, synthesis-feasible lead series.
Platform Architecture
Modular Architecture
The most contemporary and best models rapidly integrated and adaptable for each discovery task, not a single “hero model"..
Reinforcement-learning Optimization
Iterative design improves candidate molecules each cycle.
Closed-loop Molecular Optimisation
Incorporates both ADMET and regulatory QSAR constraints (NAM-aligned).
Computational Risk Elimination
Three validation layers — structural confidence, ADMET/safety, and regulatory tox — eliminate clinical failure modes before synthesis. 250+ properties per compound.
Why the Optimization Loop Matters
Competitors optimize molecules against potency and ADMET independently, then check regulatory toxicology downstream — surfacing liabilities late. CureRX embeds regulatory QSAR constraints directly into the RL optimization loop, eliminating compounds with hERG, AMES, or CYP liabilities before they consume synthesis budget.
The FDA Modernization Act 2.0 and April 2025 FDA roadmap are accelerating adoption of non-animal methods. CureRX’s in-silico ADMET and regulatory QSAR workflow is natively aligned to this shift, generating partner-grade data packages that reduce downstream animal study burden and accelerate IND-enabling timelines.
Nine lead candidates generated across three biological targets, with programmes advancing through experimental validation.
​Three synthesis-ready lead series generated in under 12 weeks for a difficult isoform-selectivity problem.
