Company Profile: Recursion

AI drug discovery is not a new phenomenon – but it is evolving at an almost dizzying pace. Recursion is one of the earliest innovators in the space, founded in 2013 when CEO and co-founder Chris Gibson first recognised the potential of using cell microscopy images to train AI algorithms.

“We had this evidence that maybe images in biology were important,” says Gibson. “And so the pitch to early investors was that, in biology, structure suits function, and if we can generate a massive library of images of biology, and if we can get really good at doing computer vision, maybe we could decode biology.”

At that time, the idea that by collecting a massive amount of machine-readable cell data AI could discover insights that human scientists couldn’t see – and then design successful drugs to go after those targets – was a massive leap of faith. Today, the value of this approach is widely recognised even by most pharmaceutical companies, a number of which – including Sanofi, Roche and Genentech, Bayer, and Merck KgaA – have partnered with Recursion to leverage the strength of the company’s platform and supercomputer, both of which have grown exponentially since those early days.

Just as AI and machine learning technologies have progressed from the ability to recognise cat pictures to the ability to detect cancerous tumours from scans, Recursion’s platform – known as the Recursion Operating System (Recursion OS) – has evolved from a point solution into a full-stack discovery and design engine that takes an AI-first approach to drug discovery and development.

They call the approach TechBio.

As opposed to traditional pharma or biotech, which begins with a hypothesis, usually based on a reductionist view of a single gene or protein target, the TechBio approach begins with data – massive amounts of data across numerous data layers collected in Recursion’s automated wet labs and, in the case of patient data, accessed via partnerships. AI and machine learning are applied to that data and reveal, via in silico modelling, new hypotheses and new drug candidates that are identified as likely to be successful. Experiments to test and validate these predictions generate even more data that further improves the platform’s predictions, resulting in a constant, iterative learning loop known as active learning.

This TechBio approach drives down cost, accelerates the speed of discovery, and improves efficiency – addressing many of the longstanding obstacles preventing more new medicines from reaching patients in a timely way. With traditional methods, it takes a decade or more and billions spent before a drug reaches the market – and 90% of drugs in development fail in clinical trials.

By leveraging data, automation, and virtual modelling, Recursion is able to develop a lead molecule with 10 times the efficiency of traditional methods by significantly reducing the number of molecules that need to be physically synthesised, in half the time and at a fraction of the cost. But beyond efficiency, the company’s platform is also capable of finding novel biology that may have been overlooked by humans, or considered too challenging to pursue. It can discover novel disease targets, precision-design highly optimised drugs, and match those drugs to the patients who are most likely to benefit.

A prime example is the company’s REC-1245 program – a small molecule drug for select solid tumours like ovarian, colorectal, and endometrial cancer that’s currently in Phase 1/2 trials.

In 2021, using an early version of the Recursion OS, scientists found that the splicing factor protein RBM39 was associated with key regulators of DNA damage response (DDR). Targeting this protein, they found, would have the same effect as inhibiting the desirable cancer target CDK12, but without the same toxic side effects.

The team next used the AI platform to design a potential first-in-class RBM39 degrader, progressing from novel discovery to lead drug in under 18 months, more than twice the speed of industry average. Preclinical studies validated the discovery. To identify the right patient population, Recursion then screened publicly available cancer patient datasets against REC-1245 and looked for relevant biomarkers, discovering a huge response, for instance, in patient models that were MSI-high, which helped guide patient selection for the clinical trial.

Recursion focuses on diseases where there is a high unmet need – including aggressive cancers and rare diseases. The company is already seeing encouraging data for some of its clinical programs, such as REC-617, a potentially best-in-class CDK7 inhibitor for advanced solid tumours that was designed with AI in less than 12 months, with just 136 novel molecules synthesised (compared to thousands required using traditional methods).

Using the Recursion OS, scientists precision-designed a molecule that tightly controls the extent and duration of target inhibition, addressing two key mechanisms – cell cycle dysregulation and transcription – that are the hallmarks of aggressive cancers while also optimising for rapid oral absorption and short half-life to minimise GI issues that are common with other CDK7 inhibitors. REC-617 is currently in an ongoing Phase 1/2 ELUCIDATE trial in patients with advanced solid tumours.

Today, human genomics data is fully integrated into the platform at every stage of the discovery and development process. Each new therapeutic programme begins by connecting data from cells to patients – finding genetic drivers for disease and matching those with the company’s unique chemistry or biology insights.

In order to fulfil its mission to “decode biology to radically improve human lives,” Recursion is continuously moving toward increased AI, automation, and machine learning modelling – maximising digital insights and reducing the need for real-world experimental validation. The company’s ultimate vision is to advance a “virtual cell” – and, someday, a virtual organ and even a virtual patient.

In May 2025, Recursion scientists released a paper outlining this vision, “Virtual Cells: Predict, Explain, Discover,” which detailed a digital cell model that can reliably simulate patient responses, allowing researchers to generate and test large numbers of therapeutic hypotheses safely and economically before initiating costly clinical trials.

While other research groups have proposed blueprints for a virtual cell, Recursion’s vision is unique in that it centres on causal models that not only predict the functional response of cells but – critically – explain those predictions. This virtual cell vision is also specifically aimed at driving therapeutic discovery – something that Recursion, with over 65 petabytes of proprietary data, automated wet labs, a robust clinical stage pipeline, and a powerful supercomputer – is uniquely positioned to do. Recursion’s AI research engine, Valence Labs, is advancing new ways to apply machine learning to biological and chemical data to drive insights – and regularly publishes and presents at leading machine learning conferences.

Recursion has released a number breakthroughs in both proprietary and open source foundation models – including Phenom-2 – the largest foundation model for cell microscopy data, trained on over 8 billion microscopy images, as well as an open source version, OpenPhenom, to benefit the broader research community; and MolPhenix, the Best Paper winner at the 2024 NeurIPS Foundation Models for Science Workshop, which predicts the effect of any given molecule and concentration pair on phenotypic cell assays and cell morphology and delivers 10x improvement over prior baselines.

In June 2025, in partnership with MIT, Recursion announced a new open source AI model called Boltz-2, the first model to combine both protein structure and – critically – binding affinity prediction. Boltz-2 approaches the accuracy of physics-based free energy perturbation (FEP) calculations while being over 1,000 times faster and less computationally expensive. The open source tool has been downloaded by over 50,000 unique users to date.

“By predicting both molecular structure and binding affinity simultaneously with unprecedented speed and scale, Boltz-2 gives R&D teams a powerful tool to triage more effectively and focus resources on the most promising compounds,” said Najat Khan, chief R&D officer and chief commercial officer at Recursion.

With each new AI tool, and each new clinical milestone, Recursion unlocks another piece of the biological puzzle, and comes one step closer to fulfilling its mission.

Recursion (NASDAQ: RXRX) is a clinical stage TechBio company leading the space by decoding biology to radically improve lives. Enabling its mission is the Recursion OS, a platform built across diverse technologies that continuously generate one of the world’s largest proprietary biological and chemical datasets. Recursion leverages sophisticated machine-learning algorithms to distill from its dataset a collection of trillions of searchable relationships across biology and chemistry unconstrained by human bias. By commanding massive experimental scale – up to millions of wet lab experiments weekly – and massive computational scale – owning and operating one of the most powerful supercomputers in the world, Recursion is uniting technology, biology and chemistry to advance the future of medicine.

Recursion is headquartered in Salt Lake City, where it is a founding member of BioHive, the Utah life sciences industry collective. Recursion also has offices in Montréal, New York, London, and the Oxford area.

Learn more at www.recursion.com, or connect on X (formerly Twitter) and LinkedIn.