Somite Becomes Cellular Intelligence, as It Builds the First Universal Virtual Cell-Signaling Model

New white paper details key technical milestones and validation of the company’s proprietary, capsule-based platform to accelerate breakthroughs across human health

BOSTON, Jan. 7, 2026 /PRNewswire/ — Somite, an AI-native TechBio company building the first universal virtual cell-signaling model, today announced its new identity, Cellular Intelligence. This rebrand reflects the company’s vision to understand, predict and control cell behavior. To mark the launch, the company released a white paper, Engineering Cell Fate: Towards a Foundation Model for Virtual Cell Signaling, detailing its approach to achieving that vision.

“Cells respond to signals in intricate ways that govern both health and disease, but we still do not understand the rules that determine these outcomes,” said Micha Breakstone, Ph.D., Co-Founder and CEO of Cellular Intelligence. “While the industry races to map the static geography of cells, we are building the navigation system to guide them. That is the shift our new name represents. By decoding the grammar of cell signaling, we are replacing trial and error with predictive engineering, allowing us to design regenerative therapies and model diseases with the precision of a true engineering discipline. We are building a future where biology is no longer destiny, but design.”

Scientific Co-Founder Olivier Pourquié, Ph.D., a tenured professor on leave from Harvard Medical School to serve as Chief Biology Officer and Head of Research at Cellular Intelligence, describes the engine behind this vision:

“Cellular Intelligence has built a proprietary capsule platform that enables cell-signaling analysis and data generation at scale, with 1,000x higher efficiency than existing methods. This enables us to test millions of signal combinations on differentiating pluripotent stem cells in parallel. This scale allows us to identify precise conditions needed to control cell differentiation and understand diseases, ultimately transforming the future of medicine.”

Engineering Cell Fate: Towards a Foundation Model for Virtual Cell Signaling
The newly released white paper outlines Cellular Intelligence’s solution to the challenge of predicting and controlling cellular behavior with the development of the first universal virtual cell-signaling model. It details how, in less than two years, the company progressed from concept to demonstrating in vitro and in silico proof-of-concept results. The white paper also outlines the company’s roadmap for the future, and highlights the applications and impact on the field, and ultimately on patient lives.

Key technical milestones highlighted in the report include:

• Unprecedented Scale: Following an early pilot screen of 27,000 unique perturbation sequences, Cellular Intelligence successfully scaled to over 1 million sequential signal combinations in 2025. This massive, context-rich dataset demonstrates the platform’s data generation capabilities and potential to cover the astronomical search space of cell signaling at a scale no other effort can approach. 
• Increased Resolution: As the data scaled, so did the granularity. For instance, where the company previously saw the derivation of “muscle precursors,” it now sees separate clusters for skeletal versus cardiac muscle lineage. This underscores that increasing data density yields increasing resolution, which will, in turn, directly strengthen the training of increasingly predictive models.
• Temporal Dynamics: Cellular Intelligence captures the dynamics of how cells change over time, rather than a series of discrete genetic snapshots, to explore the vast array of intermediate cell states produced during development. 
• Mapping Healthy Development to Decode Disease: By understanding healthy development, Cellular Intelligence can decode disease. Healthy and diseased cells both use the same set of approximately 20 core molecular signaling pathways, and it is the combinations, doses and temporal order in which they are used that drive the immense diversity of human cell types and states. 
• Active Learning and Data Augmentation: Instead of testing perturbations randomly or exhaustively, the model identifies gaps in its understanding – perhaps a particular signal’s effect on a specific cell subtype remains poorly predicted – and prioritizes those experiments, creating a self-improving engine.

Arjun Raj, Ph.D., a tenured professor on leave from the University of Pennsylvania to serve as Head of Computational Biology, details the impact of this data engine:

“We have successfully executed some of the largest combinatorial signaling experiments the field has seen, demonstrating the ability to scale to over 1 million unique combinations, and we have already proven that our platform can generate a diverse set of cell types within a single run. This leaves us at a critical inflection point: our data engine is operating at the scale required to unlock the power of machine learning techniques, with massive, high-fidelity datasets yielding more powerful and generalizable models than ever before. These advances stand to transform research and development.”

To learn more about Cellular Intelligence and read the white paper, visit www.CellularIntelligence.com.

About Cellular Intelligence
Cellular Intelligence is an AI-native TechBio company building the first universal virtual cell-signaling model to decode and control the language of cells. Using our proprietary capsule technology, we generate dynamic cellular data at over 1,000x higher efficiency than existing methods. We train large-scale predictive models on this data to transform biology into an engineering discipline, enabling us to design and optimize regenerative medicine protocols, predict drug responses, and computationally model disease mechanisms. Our founding team includes repeat AI entrepreneur Dr. Micha Breakstone (Chorus.ai, acquired for $575M), the Head of MIT’s Fundamental AI Group, and four world-renowned professors, including three Harvard Medical School professors, three members of the National Academy of Sciences or Medicine, and the Chair of Genetics at Harvard Medical School. Based in Boston, Cellular Intelligence has raised over $62M to date from Khosla Ventures, the Chan Zuckerberg Initiative, SciFi VC, AMD Ventures, Fusion Fund, and others. Learn more at www.CellularIntelligence.com.

Contact: Media-Relations@CellularIntelligence.com

View original content to download multimedia:https://www.prnewswire.com/news-releases/somite-becomes-cellular-intelligence-as-it-builds-the-first-universal-virtual-cell-signaling-model-302654726.html

SOURCE Cellular Intelligence