Product

UNMIRI Platform

UNMIRI is one technical platform with multiple product surfaces, serving the precision oncology software ecosystem at three layers: APIs for healthtech vendors, intelligence tools for biotech medical affairs, and free tooling for the clinicians who interpret cancer genomic data daily.

For healthtech and EHR vendors

Two APIs, one shared data plane.

Engineering teams shipping oncology features can plug in UNMIRI's parsing and decision-support APIs instead of rebuilding cross-vendor genomic interpretation in-house. The two APIs share a knowledge graph, a parser stack, and an output schema, so rolling out the second one against an existing integration is incremental work rather than a second project.

For biotech medical affairs

Citation-grounded oncology intelligence.

Medical affairs teams need defensible evidence faster than the incumbent tooling delivers it. Engine 3 gives oncology medical affairs teams a literature surveillance and KOL-tracking layer with citation-grounded summaries that hold up to medical leadership review, with a self-serve on-ramp that fits a mid-market biotech or an oncology medical affairs group inside a larger pharma, without a top-down enterprise purchase.

For pathologists and oncologists

A free tool for the people doing the interpretation.

Clinicians who interpret NGS reports daily are the most underserved group on this list. Engine 4 is built for them, distributed for free, and supported by the same parsing infrastructure that drives the paid APIs. There is no enterprise sale and no procurement process, so an individual pathologist can sign up, upload a report, and use the output the same day.

Additional APIs

Two more APIs on the same data plane.

The four engines above are the core platform. Two additional APIs run on the same knowledge graph and parser stack, for teams that need variant-grounded trial matching or prior-authorization decisions:

  • Trial Matching APIVariant-grounded trial matching that refuses to conflate L858R with T790M, with citations back to each trial's eligibility criteria.
  • PA Decision EngineVariant-aware, FDA-label-aligned prior-authorization decisions with the source excerpt behind every call.

Further capabilities stay on the platform roadmap until customer demand pulls them forward: specialty molecular pathology coding, hereditary cancer risk assessment API, real-world evidence data licensing, and others.

Architecture

One foundation, four products.

All four engines share a common technical foundation, designed for the precision oncology workflow rather than for general-purpose medical Q&A. Each architectural decision below is a deliberate move away from patterns that fail when clinical accuracy is the bar:

  • GraphRAG over a Neo4j knowledge graph
  • Authoritative sources (CIViC, ClinVar, ClinicalTrials.gov, openFDA, CPIC)
  • Deterministic output rendering, with narrow LLM use scoped to extraction edge cases
  • HIPAA-ready architecture on AWS under an active BAA, Microsoft Azure OpenAI for narrow LLM inference under the Microsoft BAA, US-only data residency

Technical deep-dives

Building precision oncology software?

Tell us which engine fits your stack. We'll route the conversation to the right person.