Ginkgo, Inductive Bio, and Tangible Scientific Connect AI Models with Automated Labs for Drug Discovery
Ginkgo Bioworks’ Datapoints unit, Inductive Bio, and Tangible Scientific have partnered to deliver a lab-in-the-loop workflow that ties AI-driven molecular design directly to experimental validation. The collaboration integrates predictive ADMET modeling, automated compound logistics, and rapid assay services, with the aim of reducing cycle times in small molecule drug discovery.
- The central AI component comes from Inductive Bio, whose Compass platform applies machine learning models trained on consortium-scale datasets to predict absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties. These models are designed to help chemists prioritize designs with better developability profiles before synthesis. By feeding back experimental results into the models, Inductive Bio enables iterative retraining and refinement, creating a continuous AI–experiment loop.
- Ginkgo Datapoints supports this loop by generating structured assay outputs that are formatted for direct integration into AI systems. Its ADME profiling includes microsomal stability, permeability, P450 inhibition, solubility, and optional toxicity screens, with data infrastructure optimized for high-throughput workflows. The emphasis is not only on experimental capacity but also on providing machine-readable datasets suited for model retraining.
- Tangible Scientific bridges the digital and physical sides of the cycle. Its API-enabled logistics system manages compound inventory, submission, and courier delivery to labs, while also returning metadata-rich results directly into researchers’ digital environments. This creates a closed loop in which chemists can submit virtual designs, trigger synthesis and testing, and receive validated results in days without manual coordination.
The combined setup is designed to operationalize AI-driven medicinal chemistry in practice. Traditional compound optimization often suffers from delays between computational predictions and experimental testing, making it difficult to apply AI models in real time.
See also: A Busy Day for Ginkgo Bioworks
By tightly coupling design, logistics, and lab output, the three companies aim to allow machine learning systems to function as part of an iterative discovery engine rather than as one-off analysis tools. For biopharma teams, the service is meant to provide immediate access to AI-informed compound design coupled with automated lab validation, without the need to build large-scale internal infrastructure.
Topics: AI & Digital
