Skip to content

For pharma, biotech, CROs, CDMOs, medtech, hospitals and payers

Ship drugs and devices 9–18 months faster.

One sparse Mixture-of-Experts AI platform across the full lifecycle: discovery, clinical, RWE, regulatory, manufacturing and commercial. Built compliance-first. Trained on diverse global populations. Production-grade in 8 weeks.

Request a showcaseSee the platform

Phase II statistical considerations

ART-4468 · draft

Under a two-sided alpha of 0.05[2] and power of 80%[2], powered to detect a hazard ratio of 0.65[1], 170 events are required, or 270 participants randomised one to one.

Citation [1]

Registrational study primary publication

Effect estimate and event maturity

Span match0.94

Computed by tool, not the model

Required events
170
Total randomised
270
Per arm
135

Schoenfeld, log-rank
genovant.stats.survival_size v1.2.0

  • 30–40%shorter Phase II setup
  • $8–15Msaved per asset
  • 8×faster cohort assembly
  • 60%faster value dossiers
  • 8 wksto production

The customer reality

Pharma R&D economics are broken, and AI sprawl is making it worse.

Every dollar spent integrating point tools is a dollar not spent on science. Genovant collapses the AI stack into one foundation and clears the GxP gate.

  • $2.6B

    Cost crisis

    Average cost to bring one new drug to market, over a 10–15 year cycle, with Phase II/III failure rates above 50%.

  • 70+

    AI tool sprawl

    Fragmented AI vendors per top-10 pharma. Protocol authoring, RWE, HEOR, safety and regulatory each run as a separate tool with no shared data spine.

  • 12–18 mo

    Slow trust-to-deploy

    Typical AI pilot-to-production cycle. GxP, 21 CFR Part 11, audit-trail and citation gates stall most GenAI tools in IT review.

  • <5%

    Diverse-population blind spot

    Share of clinical-trial data representing non-Western patients. Models trained on US and EU data misprice efficacy, safety and access for most of the world.

Sources: Tufts CSDD / DiMasi; McKinsey GenAI in pharma; FDA Drug Trials Snapshot; Genovant analysis of top-20 pharma AI vendor footprints.

How Genovant works

Three decisions. One asset. One cohort definition that drives all of them.

  1. 1Design · GenoDev

    Is this trial design defensible, and who else is in this space?

    Eligibility criteria in ICH M11 form. Sample size from a named calculator, with every assumption cited.

  2. 2Feasibility · GenoRWE

    How many patients actually meet these criteria, and is the data even there?

    The criteria arrive as a structured artifact, resolve to concept sets and run as an attrition funnel.

  3. 3Access · GenoAccess

    What will payers say, and what evidence do we need now?

    The eligible count flows straight into budget impact. No second spreadsheet.

Each step passes a structured, cited artifact to the next, not a chat transcript.

Cohort feasibility

Synthetic dataset

A rule, not a code

HER2 IHC = 1+ORHER2 IHC = 2+ANDHER2 ISH = negative

  1. Breast cancer, any stage128,400
  2. Unresectable or metastatic18.8%24,180
  3. HER2 status recorded at all70.0%16,935
  4. HER2-low, by rule48.5%8,214
  5. At least one prior line71.7%5,890
  6. Performance status 0 or 157.9%3,412
  7. No prior antibody–drug conjugate87.2%2,977
Nearly a third of otherwise eligible patients drop out here because the biomarker was never recorded, not because they failed it. Bar length shows retention from the previous step.

Beyond retrieval

Some criteria aren’t codes. They’re rules.

A biomarker status defined by two lab results cannot be looked up as a single concept. Genovant builds it as a rule over observations, runs it against the data and shows exactly where patients drop out.

Often the binding constraint isn’t the disease. It’s that the test result was never recorded. We show you that step too.

How GenoRWE builds cohorts

The platform

One platform, three products, built for the entire drug and device lifecycle.

Adopt one product to solve a function. Adopt the platform to compound value with every team you onboard: the same foundation, the same compliance shell, the same data spine.

  • GenoDev

    Clinical Development

    From protocol to first patient in.

    • AI protocol authoring aligned to ICH M11
    • In-silico trial simulation and site scoring
    • Biomarker and genomic integration
    • Adaptive design and feasibility forecasting
    30–40%shorter Phase II setup
    Explore GenoDev

  • GenoRWE

    Real-World Evidence

    Cohorts and prevalence in days, not quarters.

    • OMOP CDM-harmonised federated search
    • Rare-disease prevalence and burden of illness
    • Comparative effectiveness
    • Claims, EHR and registry triangulation
    8×faster cohort assembly
    Explore GenoRWE

  • GenoAccess

    Market Access & Value

    Evidence-grade payer stories.

    • HEOR cost-effectiveness modelling
    • Payer policy and pricing simulation
    • Auto-drafted value dossiers and AMCP packages
    • Competitive intelligence and HCP segmentation
    60%faster dossier first draft
    Explore GenoAccess

What we never compromise

Four things that aren’t features. They are the product.

  • The citation gate

    Nothing is shown until each claim matches its source span and passes an entailment check.

  • The intended-use boundary

    Diagnosis or treatment questions about an identifiable patient are declined, and the decision is logged.

  • The audit trail

    Every route, tool call, source and check, visible on screen and exportable.

  • Honest disagreement

    When sources conflict, both are shown and routed to a person. Never averaged.

Read our trust and compliance approach

Runs where your data is allowed to live.

Data rules are decided before any retrieval or model call. Pick the mode that fits your residency, IT and risk posture.

  • Managed SaaSManaged, for fast pilots
  • Customer VPCYour AWS, Azure or GCP account, your keys
  • On-premisesAir-gapped, for sovereign data

Engagement model

Land with a paid pilot. Expand with a platform licence.

Consumption-based pricing with predictable enterprise floors as you scale. Pilot fees credit 50% against the platform licence.

  1. Phase 1

    Pilot

    8-week paid pilot

    Fixed scope, fixed fee. One use case, one product. Success metrics co-defined and signed in week 1. Working model and ROI baseline at week 8.

    Indicative $250K–$500K

  2. Phase 2

    Expand

    12-month platform licence

    Roll out further use cases on the same tenant, across GenoDev, GenoRWE and GenoAccess. Per-seat and per-query economics.

    Indicative $1.5M–$3M / yr

  3. Phase 3

    Enterprise

    Multi-year agreement

    Full platform access, unlimited business units, preferred pricing floor locked. Custom expert training, dedicated infrastructure, named CSM.

    Indicative $3M–$8M / yr

Our commitments to you

  • No data buyout, no source-code lock-in. You own your tenant data and models.
  • Pilot pricing credits 50% against the platform licence. Floor locked at pilot graduation.
  • 30-day exit clause through the pilot. 90-day data egress when an enterprise contract ends.

Three design-partner slots open this quarter: preferred enterprise pricing locked for 24 months on pilot graduation, a seat in roadmap design and a co-published case study.

See it work on a real question.

Forty-five minutes, one asset, from trial design to reimbursement. Public and synthetic data only.

Request a showcase