ORDOS
Foundation models
for biology.
We are building the computational foundation for the next generation of medicine. It begins with OSMO, a model that predicts how a drug moves through the human body from its structure alone.
The problem
Medicine is still discovered by trial and error.
Every drug program asks the same question: what will this molecule do inside a human body. Today that answer is earned slowly, through years of experiments, animal studies, and models that specialists assemble by hand over weeks.
Most of biology is still explored one costly result at a time. The knowledge exists, but it lives in the intuition of a small number of experts and in data that no team can read all at once. The bottleneck is not effort. It is prediction.
Slow
A single pharmacokinetic model can take a specialist weeks to build and validate.
Expensive
Programs commit years and capital before they learn how a molecule behaves in the body.
Scarce
The expertise to answer these questions lives with a handful of people.
Why now
Three curves are meeting at once.
Foundation models learned to read the language of proteins and molecules. Decades of pharmacological measurement became structured, machine-readable data. Compute grew cheap enough to model the body as one connected system. For the first time, prediction can lead instead of follow.
What is OSMO
OSMO is a foundation model for pharmacokinetics.
Give it three things, and it predicts how a drug distributes across the human body.
Input 01
Molecular structure
The drug itself, described by its chemistry as a structure or a SMILES string.
Input 02
Dose
How much of the drug is given, and how often it is taken.
Input 03
Formulation
How the drug is delivered into the body, from an oral tablet to an injection.
Structure, dose, formulation whole-body prediction
In seconds, OSMO returns organ-level exposure, concentration over time, and predictions you can interpret. No manual model building. No prior laboratory measurement required to make a first prediction.
How it works
From a molecule to the whole body.
OSMO turns a chemical structure into a full picture of exposure across the body, and shows the reasoning behind it.
Stage 01 · In
The molecule
You provide the structure, the dose, and the formulation. Nothing else is required to begin.
Stage 02 · Model
OSMO
The model reads the chemistry, infers the physical properties that govern how the molecule travels through tissue, and simulates its movement through the body as one coupled system.
Stage 03 · Out
The body
You receive concentration over time, everywhere it matters.
- Plasma and major organs
- Concentration-time profiles
- Interpretable parameters
Why it matters
Prediction changes where discovery begins.
Start with an answer
Rank and compare molecules for how they will behave in the body before a single experiment is run.
Fail earlier, cheaper
See exposure problems in seconds, when a program can still change course, rather than quarters later.
Make expertise scale
The judgment that once lived with a few specialists becomes available to every team, on demand.
The vision
Computational prediction, at the start of every drug program.
OSMO is the first model in a larger system. Over time, ORDOS is building toward a predictive account of human biology itself, a way to ask how a molecule will behave in bodies, conditions, and populations that no experiment has yet reached. A new computational layer, underneath all of drug development.
