Cellular Intelligence Puts AI World Models Behind a Parkinson’s Therapy
Cellular Intelligence combines predictive biology, STEM-PD, and a high-profile scientific advisory board

BOSTON — Cellular Intelligence is applying artificial intelligence world models directly to clinical cell therapy, using predictive machine-learning architectures in an experimental Parkinson’s disease treatment while building an advisory board that spans deep learning, biotechnology, tissue engineering, and computational biology.
Founded in 2023 as Somite.ai, the Boston-based company is attempting to replace decades of manual, trial-and-error laboratory research with computational systems that predict and direct cellular behavior. Its models are designed to learn how cells process signals and change states over time, rather than relying only on high-throughput screening of chemical compounds or living cells.
Cellular Intelligence has raised more than $70 million in total capital. Its investors include Khosla Ventures, the Chan Zuckerberg Initiative (CZI), and Novo Nordisk, which took an equity stake in the company as part of a transaction involving the company’s clinical program.
That program is STEM-PD, a cell therapy for Parkinson’s disease that Cellular Intelligence acquired global rights to in May. Novo Nordisk had shuttered its internal cell-therapy R&D unit in October during a broader corporate realignment toward metabolic treatments. Under the acquisition terms, Novo Nordisk remains eligible for future development and commercial milestones, along with sales royalties.
Parkinson’s disease is a progressive neurodegenerative disorder in which dopamine-producing neurons in the brain are lost. The resulting motor symptoms include tremors, rigidity, and bradykinesia. STEM-PD uses stem-cell-derived dopaminergic progenitor cells intended to replace lost neuronal tissue after transplantation.
The therapy has Investigational New Drug (IND) clearance from the U.S. Food and Drug Administration (FDA) and FDA Fast Track designation. Fast Track provides an expedited regulatory review pathway for therapies addressing serious conditions with unmet medical needs. STEM-PD is now Phase 2-ready and is Cellular Intelligence’s sole owned clinical-stage asset.
“Novo Nordisk selected CI to advance STEM-PD, and we acquired global rights to a Phase 2-ready Parkinson’s program with FDA IND clearance and Fast Track designation,” said Micha Breakstone, chief executive officer of Cellular Intelligence. “Novo also made a strategic equity investment in CI. We see that as a strong vote of confidence in the next chapter of this work.”
The company’s commercial and scientific strategy has two parts: developing proprietary clinical therapeutics and licensing its biological foundation models to external pharmaceutical companies for target discovery and development. Cellular Intelligence plans to use its capital and advisory structure to advance the STEM-PD clinical trial while pursuing external commercial partnerships around the foundation software models.
Those models rely on what artificial intelligence researchers call “world models,” algorithmic frameworks that map the underlying rules of a physical or biological system. The software can then simulate and forecast how that system will respond when acted upon.
In the company’s pipeline, the system tracks temporal biological data, including how living cells respond over time to external inputs, molecular signals, and environmental changes. It combines datasets from single-cell genomics, developmental biology, and machine learning to forecast cellular differentiation pathways without requiring exhaustive physical testing for every condition.
According to internal benchmark data cited by Cellular Intelligence, its foundation model has shown stronger predictive accuracy than existing computational tools when modeling normal developmental processes and oncological pathways. The model was not trained directly on cancer cell datasets.
“A world model learns how a system behaves and predicts what happens when you act on it,” Breakstone said. “For us, the system is a living cell: the basic computational unit of life, receiving signals, processing information and changing its behavior in response.”
The scientific advisory board announced Monday includes Yann LeCun, who shared the 2018 ACM A.M. Turing Award with Geoffrey Hinton and Yoshua Bengio. The award is often described as the Nobel Prize of computing. LeCun is a pioneer in deep learning and convolutional neural networks and led Meta’s Fundamental AI Research (FAIR) lab as chief AI scientist from 2013 until December.
LeCun left Meta in December to become executive chairman of AMI Labs, an AI venture focused on world-model architectures. Another new adviser, Robert S. “Bob” Langer, is an Institute Professor at the Massachusetts Institute of Technology (MIT), the university’s highest faculty honor, and is also serving as an observer on Cellular Intelligence’s board of directors.
Langer received the National Medal of Science and holds more than 1,500 granted and pending patents in tissue engineering and drug delivery systems. He co-founded Moderna in 2010 alongside roughly 40 other biotechnology startups. His recent venture activity has included Vivtex, an oral-drug delivery startup that entered a $2.1 billion licensing deal with Novo Nordisk.
Langer is also connected to T.Rx Capital, a venture firm founded by his son Michael Langer, and Soufflé Therapeutics, an RNA startup co-founded by his daughter Susan Langer.
“What matters to me is whether the science can lead to something that makes a difference for patients, and whether the people involved can carry it through,” Langer said. “I have been doing this for over 50 years, and Micha and the scientific team he has brought together are among the very best I have seen.”
Langer noted that Cellular Intelligence combines a proprietary temporal biological data engine, a specialized foundation model, and an in-house clinical pipeline to test predictions in real-world settings.
Fabian Theis and Jens Nielsen have also joined the advisory council. Theis is a computational biologist who directs the Computational Health Center at Helmholtz Munich. He co-founded the Human Cell Atlas, an international consortium seeking to map all human cell types through single-cell genomics, and received Germany’s Gottfried Wilhelm Leibniz Prize in 2023 for machine-learning methods in cell biology.
Nielsen is chief executive officer of the BioInnovation Institute (BII) in Copenhagen. BII is an incubator created by the Novo Nordisk Foundation to scale life-science startups. Fidji Simo, the chief executive of applications at OpenAI and former CEO of Instacart, is additionally serving as a strategic advisor to Breakstone on corporate execution.











