Optagon Labs
Optagon Labs provides an AI-native R&D service that models the entire drug development program—molecule, formulation, process, and body—as a single digital twin. Its frontier model co-optimizes coupled drug properties and selects only the highest-value experiments to run, helping biotechs converge on clinical candidates faster. The system integrates a molecular foundation model, state representation architecture, and sequential optimizer with interpretable, provenance-tracked decisions.
- Drug Discovery & Therapeutics
- Artificial Intelligence
- Biotechnology
- AI Agents
- Software Only
- Healthcare Technology
Funding
Founders
Founder details are not available yet.
Product
Problem
Drug engineering—the process of optimizing a molecule for solubility, stability, safety, permeability, and manufacturability—remains a bottleneck in pharmaceutical development, often consuming 5+ years and tens of millions of dollars. Teams of up to 300 scientists run siloed, manual experiments across multiple stages (hit-to-lead, lead optimization, candidate selection, early CMC), and design decisions made early can lock in constraints that surface as toxicity, poor exposure, or formulation failures in clinical trials. Most biotechs lack the AI infrastructure to address this multi-variable, sequential search problem under high uncertainty.
Solution
Optagon Labs provides an AI-native R&D service that brings the entire drug development program into one digital twin, modeling the molecule, formulation, process, and the body simultaneously. The system co-optimizes multiple drug properties at once by running design–make–test loops across internal and external labs, selecting only the high-value experiments that resolve the most uncertainty. Its frontier model comprises three integrated components: a molecular foundation model, a state representation architecture, and an optimizer that decides which experiment to run next. Every decision is interpretable and carries its uncertainty and provenance, allowing drug developers to converge rapidly on optimal designs that satisfy potency, ADMET, and CMC targets before entering clinical trials.
Target Audience
Primary customers are biotech companies and pharmaceutical firms that need to engineer drug leads into clinical candidates faster and with lower risk, particularly small biotechs operating on finite runway that cannot build their own frontier AI capabilities.
Features
- Digital twin that simultaneously models the drug lead's molecule, formulation, process, and physiological environment across all four development stages
- Co-optimization of coupled parameters (e.g., core scaffold, logD, pKa, salt form, particle size, crystallisation solvent) against targets like potency, hERG margin, oral exposure, and impurity profile
- AI-driven experiment selection that runs only the tests that maximally reduce program-level uncertainty, avoiding wasteful trial-and-error
- Frontier model integrating a molecular foundation model, a state representation architecture, and a sequential optimizer—not siloed single-endpoint predictors
- Interpretable decisions with provenance tracking, enabling traceability from design choices to clinical outcomes
- Human-in-command orchestration with modular AI accelerators that share context across different R&D endpoints, versus fully autonomous black-box systems
- Modular deployment of agents at silos close to verification, batching long-horizon thinking into connected short-horizon intelligence buckets with a transcendent ontology