—
—
Most blockbusters aren't new molecules. They're new schedules.
The whole organism, as one program. We discover dose schedules, drug combinations, and repurposing candidates against coupled human physiology — for pharma, biotech, and academic research.
NOT AN LLM. A composed system of ordinary differential equations, parameterised from peer-reviewed studies, solved end-to-end using proprietary computational techniques rooted in quantitative finance. Every output is a reproducible numerical solution — not generated text.
→ Click any node in the atlas to see what it models. Scroll to see how each capability uses a different chain.
Compatible with how pharma already buys this.
The mechanistic-modelling tradition has a 40-year track record at the EMA and FDA. Our work sits inside that tradition and uses its vocabulary. The distinctive layer is the proprietary computational stack that lets the whole composed body run end-to-end at sub-second wall-clock per scenario — which is what makes sweep, counterfactual, and combination-search workflows practical rather than aspirational.
The dose is the model output, not the model input.
Given a tumour model coupled to bone marrow, plasma drug kinetics, and hepatic clearance, we search the schedule — daily dose, holiday weeks, ramp-up, intermittent dosing — that maximises response under a marrow-toxicity constraint. The chain on the right is the minimum required to compute it.
We expose: daily dose, drug-holiday length, ramp profile, max cumulative dose. The optimiser balances tumour AUC vs neutrophil nadir vs hepatic enzyme rise. Wall-clock per schedule sweep: seconds, not days.
Two drugs, one body — predicted synergy and shared toxicity, in the same pass.
Combination response is rarely additive. Our composed body computes the cross product of approved drugs against a target indication, scoring each pair on response surface (Bliss / Loewe) and on shared toxicity routes — most commonly competitive CYP inhibition and overlapping marrow suppression.
Surfaces a small set of high-synergy / low-shared-tox candidate pairs out of the 100-million-pair search space. Pre-filters before any wet-lab confirmation.
Look at the mechanism. Find drugs whose mechanism happens to hit it.
Pick an under-served indication. We score every approved compound by how its mechanism (target affinities, off-target hallmark profile, tissue penetration) matches the indication's hallmark signature, subtracting toxicity through shared substrates.
Output: a ranked shortlist of plausible repurposing candidates with a defensible mechanism story, ready for a focused in-vitro panel.
Replicate the published trial. Then extend to the patients the original sponsor did not study.
For any published Phase II / III, we generate a virtual cohort matched to the trial's demographics, apply the same dosing protocol, and check that endpoint distributions land within the reported confidence intervals. Then we change the cohort and recompute.
39 published trials replicated. Useful for label-extension scoping, off-label safety, sub-population dose adjustment.
What would the AE rate have looked like if drug A's CYP inhibition had been flagged at IND?
A drug-induced cardiac event in a real trial usually reflects an interaction — the indexed drug shifts hepatic CYP3A4 activity, a comedication's AUC rises, QTc prolongs, the patient flags TdP. Our model runs the counterfactual.
Used for retrospective AE attribution, prospective IND CYP-screen prioritisation, label-language scoping.
Formulation, route, depot — scored on what reaches the target.
Oral, SC, IV, IM, depot, inhaled, intranasal — every route lands on a different gut / hepatic / lymphatic / nasal substrate, and the bioavailability into the target organ depends on the chain.
Surfaces formulation re-ranking, route trade-offs, depot-vs-daily, BBB-penetration likelihood.
One body, one model, one conversation. Bring the question.
Pharma, biotech, academic research. We start with a worked example on your published case, then scale.