Data, science, and outcomes - behind Roxi-Ai

We are building the first longitudinal, outcome-linked risk signal platform for biopharma- so investors and leaders can act earlier, de-risk systematically, advance winners, and accelerate the development of life-saving medicines.

Programs & ecosystem

Member, NVIDIA Inception Program

Member, AWS Activate – startups

Our story & mission

Great scientists, clinicians, and investors make hard calls every day—but the decisive signals they need are scattered across silos. Roxi-AI unifies that evidence and turns it into explainable, outcome-linked risk signals your teams can trust.

Our mission is to help people make faster, fairer, and more confident decisions—so capital reaches the right assets earlier and patients receive better therapies sooner.

What we do?

Proprietary risk data foundation

Roxi-Ai unifies messy clinical, scientific, regulatory and market evidence into a knowledge graph designed to surface success/failure signals years before outcomes are known.

Explainable predictive models

Our Roxi-100 framework codifies 900+ weighted, time-stamped risk signals linked to outcomes, so teams can see why risk shifts—by domain, over time, and versus peers.

Actionable readouts & scenarios

Tools include the predictive Roxi IndexBenchmark Pro for peer comps, the Simulator for portfolio/scenario stress tests, and Ask Roxi, your interactive diligence agent.

 Validation

92% directional accuracy at Phase-1 start on a blinded, retrospective oncology cohort (n=64) – with clear early signal separation between eventual approvals and failures. Full methods, assumptions, and limitations available on request.

Meet Our Leadership Team

Deepak Patil, MBBS, MBA, MS

Founder

Lorraine Marchand, MBA

Founder

Partner / Contact

We partner with hedge funds, VCs/PE, and biopharma teams on diligence sprints, portfolio/asset reviews, and search-and-evaluation workflows- blending standardized risk readouts with explainable drivers and peer benchmarks. Book time to scope your use case, or start with a ticker/asset readout.