Key Takeaways

  • Biohub began with an audacious goal from Mark Zuckerberg: to “cure, prevent and manage all disease by the end of the century,” a vision Priscilla Chan admits was met with laughter by Nobel-winning scientists.
  • The mission quickly pivoted from directly curing diseases to building powerful, open-source tools that accelerate the entire scientific field, rather than tackling every disease head-on.
  • Their core strategy hinges on coupling frontier AI labs with frontier biology efforts. This isn't just about applying existing AI; it's about generating novel biological data specifically to train new AI models, pushing biology from discovery to an engineering discipline.
  • Alex Rives emphasizes the need for a new institutional model where frontier AI and frontier biology are in constant feedback, allowing models to learn directly from living systems.
  • This ambitious approach relies on Biohub's Hierarchical Approach to Biological Systems Modeling to systematically understand biology, starting from its most basic building blocks.

The Biohub's Hierarchical Approach to Biological Systems Modeling

Biohub recognizes that truly understanding and eventually engineering biological systems requires a structured, multi-level modeling effort. You can't just jump to the big picture; you have to build understanding from the ground up, ensuring a complete description at each stage.

  • Level 1: Understand Protein Interactions:

You need to be able to understand the protein interactions in order to be able to understand how cells work. So you can't just go straight to cells in a way without understanding the protein modeling.

  • Level 2: Understand How Cells Work:

If you're trying to understand something like the way the immune system works or a bunch of cells interact together, then it's tough to do that without first understanding cells. You kind of want to build the simulations at each level hierarchically, starting with the building blocks and the protein.

  • Level 3: Understand Complex Systems (e.g., Immune System, Multi-Cell Interactions):

You can understand complex systems by building up simulations hierarchically, ensuring a complete description and models that generalize by having the right basis for modeling at every level.

When This Works (and When It Doesn't)

This hierarchical approach is essential “if you really want to understand how a biological system is going to work,” as the Biohub team suggests, starting with basic building blocks like proteins. It’s ideal for ambitious, long-term endeavors aimed at achieving a fundamental, comprehensive understanding of a complex system, not just a surface-level fix. It's how you move from observation to engineering, from "what happens" to “how to make it happen.”

However, this deep, multi-level modeling requires immense resources, specialized talent across disciplines (AI, biology, engineering), and significant time. It's not a framework for quick, heuristic solutions or for problems where a 'good enough' answer is acceptable. For a founder racing to a Minimum Viable Product, attempting a full hierarchical simulation might be overkill. It also breaks down if the underlying "building blocks" at Level 1 are poorly defined or unobservable. If your system's fundamental interactions are unknown or too noisy, you'll struggle to build reliable models further up the chain.

What to Do With This

As a founder building a complex product or system, you might not be curing diseases, but you are building a biological system of a different kind—an organization, a product, a market. Apply Biohub's hierarchical modeling to your own domain this week. Pick a critical, complex system within your business, like customer churn or onboarding, and break it down:

1. Level 1 (Protein Interactions): Identify Atomic Actions. What are the smallest, discrete actions or data points? For customer churn, this might be a single login, a feature click, a support ticket, or a payment failure. Don't aggregate yet; get to the rawest signal.

2. Level 2 (How Cells Work): Understand Module Behavior. How do these atomic actions combine into meaningful user behaviors or product interactions? Does a sequence of feature clicks lead to feature adoption? Does a lack of logins and no support tickets indicate disengagement? Model these 'cellular' units of behavior.

3. Level 3 (Complex Systems): Model Ecosystem Interactions. How do these modules interact to form the overall system you're trying to understand or optimize? How does product usage, support interactions, and billing status combine to predict overall customer lifetime value or churn risk? Building your simulation hierarchically like this, rather than trying to model everything at once, will reveal the true leverage points and prevent you from mistaking symptoms for root causes. Your goal is not just to observe churn, but to engineer its prevention by understanding its underlying mechanisms.