Recursion Pharmaceuticals

Definition

Recursion Pharmaceuticals is a clinical-stage biotechnology company that uses automated cell biology and machine learning to discover new drug candidates, generating large volumes of cellular imaging and genomic data through robotic laboratory systems and analyzing them with computer vision and other models to identify relationships between diseases, genes, and chemical compounds. It positions its combined data-generation and machine learning infrastructure, which it calls a discovery platform, as a way to search a wider space of biological hypotheses faster than conventional hypothesis-driven research.

Overview

Recursion Pharmaceuticals was founded on the idea that biology could be studied more like a large-scale search problem: instead of a scientist forming one hypothesis at a time and testing it, automated systems could run millions of cellular experiments in parallel and let machine learning models find patterns a human researcher would be unlikely to notice or prioritize. This approach targets the same fundamental problem facing all drug developers — most candidate therapies fail, often because the underlying biology was misunderstood — but attacks it through breadth of experimentation rather than depth of manual investigation. Operationally, Recursion runs robotic laboratories that expose human cells to genetic perturbations, disease-relevant conditions, or chemical compounds, then photograph the cells using high-content microscopy imaging. These images are processed by computer vision models trained to detect subtle morphological changes invisible to the human eye, effectively turning cell images into a quantitative fingerprint of the cell's biological state. By comparing these fingerprints across experiments, the platform can infer whether a disease-causing gene mutation produces a similar cellular signature to a known drug, suggesting the drug or a similar compound might treat the disease — a technique broadly known as phenomics. Within the landscape of AI-driven biotech, Recursion is notable for the sheer industrial scale of its wet-lab operations, which it has marketed as one of the largest biological and chemical datasets of its kind. This distinguishes it from companies that emphasize protein structure prediction or molecule generation as their core technology, and from smaller computational-only drug discovery firms that rely on third-party or public data. Recursion has also pursued mergers and partnerships, including combining with Exscientia, another AI drug discovery company, to broaden its computational chemistry capabilities alongside its phenomics strength. In practice, the platform is used both to advance Recursion's own internal pipeline of drug candidates into clinical trials and to support collaborations with larger pharmaceutical partners who license access to specific disease datasets or the platform's predictive capabilities. Recursion has taken several of its own candidates into human clinical trials, giving it a track record, positive or negative, that purely preclinical AI discovery companies lack. The approach has clear trade-offs. Cellular imaging signatures are correlative, not mechanistic, so a drug and a disease sharing a similar image fingerprint does not guarantee a causal therapeutic relationship, and false leads are common. Clinical trial failure rates for AI-discovered candidates have not been dramatically different from industry baselines so far, underscoring that faster hypothesis generation does not eliminate the biological and regulatory uncertainty inherent in developing any new medicine.

Key Features

• Robotic laboratories running millions of parallel cellular experiments
• High-content microscopy imaging analyzed by computer vision models
• Phenomics approach comparing cellular 'fingerprints' across conditions
• Maintains one of the largest proprietary biological imaging datasets
• Merged with Exscientia to add computational chemistry capabilities
• Advances both internal drug candidates and pharma partnerships
• Multiple internally discovered candidates reaching clinical trials

Use Cases

Alternatives

Frequently Asked Questions

Related in AI Tools