Science.
Research infra, instrumentation, and talent — discovery as a fragile network.
Scientific output depends on a thin layer of unique instruments, facilities, and people. Most of these dependencies are invisible to funders and program managers — and to the institutions that depend on them. The cryo-EM facility, the synchrotron beamline, the principal investigator with the technique nobody has reproduced.
When a key node fails — a facility goes down, a key PI leaves, a grant cycle ends without renewal — the cascade isn't visible until grants stall, papers stop landing, and trainees flow elsewhere. Often that's a year of lost productivity that wasn't accounted for in any portfolio review.
Funders and institutions both run portfolios. Neither runs a graph. The dependency between bets — shared equipment, shared cores, shared mentors, shared cohorts — gets glossed over until something fails inside it.
Manifold ingests the research-infrastructure graph (facilities, instruments, key personnel, training pipelines) and the funding-portfolio layer above (grants, project pipelines, milestone obligations). Edges encode dependency: which projects depend on which instruments, which trainees on which mentors, which programs on which cores.
PEARL runs counterfactuals on facility outages and personnel departures. PARETO simulates the resulting productivity cascade across the portfolio. SPIRTES surfaces dependencies that publication data implies but org charts don't.
Institutional output depends on a thin layer of unique instruments and people, and that dependency isn't tracked. When a key facility goes dark or a key PI leaves, the cascade isn't visible until grants stall.
ΩF across the research graph; restoration latency on key nodes. Defensible succession and capex case at the institutional level.
Portfolio review is qualitative. The structural-fragility view — which instruments, which collaborations, which talent pipelines are decisive for the portfolio's deliverables — is invisible.
Per-portfolio ΩF. Identification of decisive instruments, facilities, and people. Defensible reallocation case.
Research bets assume inputs are fungible. Often they aren't — a single instrument supplier, a single CRO, a single specialist team can stall a program for a year.
Graph view of the research supply chain. Cascade simulation on supplier and CRO failure across the pipeline.
Run Manifold on a science graph.
Trial accounts come pre-loaded with a curated dataset. Or request an invite to bring your own graph.
