Healthcare OS
The EMR knows everything and explains nothing.
- EMR
- Encounters, orders, results, notes
- Scheduling
- Capacity, appointments, theatre time
- Billing and claims
- Charges, denials, payments
- Supply and pharmacy
- Stock, administration, cost
- Workforce
- Rosters, skills, hours
The record is complete. The reasoning across records is done by people, in meetings, after the fact.
Plate data
- Industry
- Healthcare
- Status
- Reference architecture
- Agents
- 05
- Layers
- 07
The architecture applied, published in full, ready to be built on.
Four questions those systems cannot answer.
- Why did length of stay rise on this ward this month?
- Which discharges will be delayed next week?
- What should we change about tomorrow's list?
- Which system carries the change, and who approves it clinically?
The industry, in miniature.
Patient, encounter, pathway, diagnosis, procedure, order, clinician, ward, capacity, claim, denial.
Ontology fragment · healthcare
ENTITY encounter
RELATIONSHIP encounter -> follows -> pathway
EVENT discharge_delayed
METRIC length_of_stay
RULE clinical_decision_boundary (enforcement: block)
DECISION discharge_planning
ACTION scheduling_updateThe blocking rule matters most here. Clinical decisions are not delegated. Operational ones can be.
Each one introduced by the question it answers.
Clinical operations
Why did length of stay rise on this ward this month?
Patient flow
Which discharges will be delayed next week, and what is blocking them?
Hospital operations
Where is capacity being lost between scheduling and theatre?
Care management
Which patients on this pathway are drifting from it, and why?
Revenue cycle
Which denials are systematic rather than individual?
What the Industry Pack contains.
- Ontology
- Encounters, pathways and capacity.
- Reference workflows
- Discharge planning, theatre scheduling and denial management.
- Agent templates
- Five agents with clinical boundaries pre declared.
- Connectors
- EMR, scheduling, billing and workforce systems.
- Evaluations
- Including cases that must fail safely.
The strictest boundary of the five.
Clinical judgment is never automated, and the architecture states that in the manifest rather than in a disclaimer.
requires_human_approval: clinical_recommendations patient_communications care_plan_changes staffing_changesOperational recommendations are produced freely. Clinical ones are produced only as evidence for a clinician, never as an instruction.
Read next.
- Pharma OS
- The one with a product behind it. Same band order, plus deployment.
- Real Estate OS
- The same seven bands, applied to a different industry.
- Manufacturing OS
- The same seven bands, applied to a different industry.
- Financial Services OS
- The same seven bands, applied to a different industry.
- The Super Brain spec
- The twelve required services and the three conformance levels.
- Industry Ontology Schema
- How the ontology fragment on this page is declared.
One architecture. Twenty pages. Same document.
This is the architecture applied, not a product yet.
Published in full and ready to be built on. If you are building it, or want us to, start with a conversation.
A conversation, not a demo script. We ask what your systems of record are, what they cannot answer, and which decisions actually matter. If an Industry OS is not the right answer for you, we will say so.