The clinician copilot

AI Hospital Guide.

Ask a guideline question in plain language and watch the answer light up on the decision chart itself. Every step is cited back to the guideline page it came from, and the clinician decides.

AI Hospital Guide

Breast · adjuvant triage · step 4 of 5

Active guidance
Workup completeStage IIAHR+ · HER2−?
Genomic assayEndocrine ± chemo
Triple negativeChemo → surgery

62-year-old, stage IIA, hormone-receptor positive, HER2 negative, after lumpectomy. What comes next?

spoken · for J.O.

Adjuvant endocrine therapy. A genomic assay decides whether chemotherapy is added; radiation follows breast-conserving surgery.

3 steps citedBreast · adjuvant triage

Cited, then decided: every step resolves to a guideline page. The clinician makes the call.

What the Guide does

From the question to the cited path, in one screen

Ask in plain language

Type or speak the question. It is routed to the right guideline of the five covered, even if you started on another cancer.

The path lights up

The cited steps highlight on the actual decision flowchart, revealed one step at a time, so you see the reasoning and not just the conclusion.

Every answer is cited

Each claim resolves to the nodes and edges of the guideline graph and the page they came from. Nothing is asserted without a source.

Starts from the patient

Pick a patient from the hospital roster and the Guide asks the guideline question with their diagnosis, stage and history already attached.

Patients · today

From the hospital record over FHIR R4

4 on panel
JO

J. Okafor

Breast · stage IIA · HR+ HER2−

Breast
MR

M. Reyes

Colon · metastatic · dMMR/MSI-H

Colon
AT

A. Tanaka

Lung · NSCLC · stage IIIA

NSCLC
DN

D. Novak

Prostate · high risk · localized

Prostate

Selected: J. Okafor. The Guide switches to the breast guideline and asks the next-step question with her stage and receptors attached.

How it is built

Built on the guidelines, wired to the hospital

Guidelines authored by hand

Each page of the NCCN Clinical Practice Guidelines in Oncology is drawn as a decision flowchart by a person, not extracted by a model. Every node keeps its link to the page it came from.

A knowledge graph, not a document dump

The flowcharts load into a GraphRAG knowledge graph. Specific questions use local search over the graph; thematic ones use community summaries. Both cite structurally.

Wired to the hospital

Reads the patient roster over FHIR R4 from InterSystems TrakCare and IRIS for Health, with OAuth or basic authentication. Without a record system it runs on a demo panel.

Decision support for clinicians. The Guide cites; the clinician decides. Guideline content is not reproduced on this site.

Guidelines covered

5 guidelines · 106 pages
  • Non-small cell lung26 pages
  • Breast24 pages
  • Testicular21 pages
  • Prostate18 pages
  • Colon17 pages

One page, one chart: each guideline page is drawn as a decision flowchart, and the same step on two pages resolves to one node.

Work with GalenOps

Run your next oncology study with agents

Book a demo and we will walk the review queue, the reasoning trail and the agent pipeline on a live synthetic Phase 3 study.

We will send a calendar link, not a newsletter.