Anto

Over 1 billion people take drugs whose success or failure is dictated by the gut microbiome, yet pharmaceutical development largely treats the microbiome as noise rather than signal. Gut bacteria can metabolize drugs before they reach targets, convert inactive compounds into active ones, transform drugs into toxic metabolites, or modulate immune tolerance, causing drugs to work in some patients and fail catastrophically in others. Clinical trials do not account for microbiome diversity across populations, so effective treatments can be abandoned as non-universal when the real driver is microbiome variability. Traditional computational methods struggle because microbiome data involves thousands of species, millions of genes, and trillions of interactions, and is roughly 99% noise, making it too sparse, high-dimensional, and confounded for reliable machine learning predictions.

Anto develops sparsification and tokenization methods designed to remove noise from microbiome data, aiming to isolate the roughly 1% of signal that carries predictive power and make the microbiome computationally tractable. The company builds foundation models for microbial communities, using causal modeling to forecast ecosystem dynamics, predict counterfactual outcomes of interventions, and enable inverse design of microbial therapies. This is paired with multi-omics data and what the company describes as 100,000+ interventional trajectories plus AI-guided lab-in-the-loop validation. Its Darwin model series is positioned to predict drug toxicity and efficacy across diverse populations and to optimize molecules for broader efficacy, targeting microbiome-driven causes of drug response and failure.