A scientific blueprint for the digital organism connects foundation models to biological simulations from molecules to the whole organism

Just as aerospace engineers test aircraft designs through computational models and simulations thousands of times before physical construction begins, a research team is proposing a scientific blueprint to give biologists the same analytical and simulation capabilities when working with biological systems. The blueprint, published on August 13, 2026 under the title "How to Build an AI-Driven Digital Organism", is led by researchers from Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in collaboration with researchers affiliated with academic institutions including the Weizmann Institute of Science, Carnegie Mellon University, and the Gene Bio AI initiative. Co-authors of the framework include Eric Xing, university president and professor of artificial intelligence, Eran Segal, dean of biological and life sciences, and Dasong Li, professor of machine learning at the university.

The scientific core of the proposal rests on establishing what is known as an "AI-driven digital organism", an integrated system of multiscale foundation models. The system relies on a modular, connectable, and end-to-end integrated design that mirrors the hierarchy of biological scales, complexity levels, and interconnectedness in living systems. The technical goal is to link specialized foundation models spanning DNA, RNA, proteins, cells, tissues, and the whole-organism level within a single continuous simulator, allowing any modification made at one level to dynamically propagate its effects across all other levels of the system.

Integrated digital simulation and targeted laboratory experiments

This digital organism serves as a safe, cost-effective alternative platform that offers high research throughput, enabling the prediction, simulation, and programming of biological phenomena across all scales, from fine molecular structures through cellular architectures to individual organisms. The computational environment allows researchers to test hypotheses and the behavior of biological pathways digitally before moving to wet-lab experiments, reducing the operational complexity and costs associated with direct experimentation in the early exploratory phases.

The research team believes that building this digital organism could initiate a more targeted and precise wave of wet-lab experiments, while advancing first-principles and data-driven scientific reasoning to help decode, interpret, and optimize life mechanisms on rigorous scientific foundations. At the same time, defining direct application areas in fields such as medicine or agriculture remains open, pending the outcomes of upcoming research and development phases.