Rethinking Medicine Through Biomedical AI

Challenge

We’ve seen medicine become more and more specialized over time. Many patients with complex conditions end up seeing five or six different specialists, and treatments are often designed for narrower and narrower subtypes of disease. But what keeps standing out to us is that biology doesn’t really fit into those boundaries.

Our Approach

This is why we’re building AI that sees the body as one connected system. It’s a lot like what happened with modern large language models: the more broadly the model was trained, the more powerful it became.

The Data Gap in Biomedical AI

AI needs a huge amount of data. ChatGPT was trained on the entire internet and all human written language, but biology isn't there yet. As a civilization, we have DNA data for fewer than five million people, and far less when it comes to RNA and multi-omics.

Internet-scale language data Billions of documents

Human DNA sequencing data < 5 million people

RNA + Multi-omics Even smaller datasets

This is one important ‘why’ we’re going global. By expanding across the US, China, India, and the Middle East, we are collecting the diverse samples needed to create a dataset that actually looks like humanity.

We’re excited to continue advancing toward a future where living to 90 is not exceptional, but expected.