One Framework.
Many Domains.

A pluggable diagnostic analytics platform
with AI as the domain expert —
and nuMetrix as the first proof.

Never just hospitals.
Always a framework.

From the start, the idea was an application framework
with a strong processing pipeline, a rich frontend,
and AI playing the role of subject matter expert.

The hospital was the first domain — not the only one.

Framework + Domain + AI

The Engine
Pipeline. Compilers. Contracts.
Taxonomy engine. Explorer shell.
Multi-tenant isolation.

How to ingest, validate, diagnose, and present.
The Domain
Entity schemas. Probe rules.
Source-system mappings.
Interpretation. Regulatory context.

What to look for and why it matters.
The AI
Subject matter expertise on demand.
Data formats. Business logic.
Medical knowledge. Regulations.

Who brings the domain knowledge — at scale.

nuMetrix — hospitals and material flow

5
tenants
including real SZO data
3
source systems
OPALE · SAP · Navision
27
probes + assessments
compiled from YAML
4
pyramid layers
probes → hypotheses → diagnoses

A full diagnostic pipeline for hospital material flow analytics.
Built on the jinflow framework. The first instantiation, not the last.

Engine vs. Domain

80% engine. 20% domain knowledge. By design.

The engine doesn't know what a “case” is.
The domain pack does.

Engine (80%)
Medallion pipeline (Bronze → Silver → Gold)
Probe compiler (YAML → SQL)
Hypothesis + Diagnosis compilers
Findings contract (10 standard columns)
Taxonomy engine (tree + closure table)
Explorer (probe discovery, interpretation, i18n)
Multi-tenant platform layer
Orchestration (Dagster, rebuild scripts)

Knows how. Knows nothing about why.
Domain Pack (20%)
Gold contract (entity schemas)
Source-system dispatch (column mappings)
Probe YAML definitions (business rules)
Hypothesis + Diagnosis YAMLs
Registry (tri-lingual display text)
Interpretation rules (context overrides)
i18n labels
Synthetic data profiles

Knows why. All declarative. All swappable.

A directory. YAML and JSON. No code.

domains/
hospital-materialflow/
contract.json
entity schema
+
probes/*.yaml
business rules
+
source_systems/
column mappings
+
i18n + registry
human language

Self-contained. Versioned. Declarative.
The engine discovers and loads it via a path variable.
Swap the directory → swap the domain.

The patterns are universal

  • Balance probes — compare two aggregates. Usage vs billing in hospitals. Received vs sold in retail. Ordered vs delivered in construction.
  • Duplicate detection — same shape everywhere. Duplicate billings, duplicate invoices, duplicate shipments.
  • Trend analysis — rolling average drift. Material costs, lead times, defect rates. The math is identical.
  • Mandatory item — “this entity must have that child.” Implant with hip replacement. Fire extinguisher with building inspection. Safety sheet with chemical batch.
  • Taxonomy hierarchies — cost centres, product categories, regional structures. Same closure table, different labels.

AI as Subject Matter Expert

The framework is the engineering. AI is the domain knowledge.

nuMetrix was built with AI
as the domain expert.

ERP
Data Formats
OPALE CSV schemas. SAP MM table structures (AUFK, MARA, EKPO). Navision column conventions. Three source systems mapped without a consultant.
Ops
Operational Knowledge
Material flow logic. Movement types (TRANSFER, ISSUE, RETURN). DRG billing rules. I/O coefficients. Ward vs OR documentation practices.
Med
Medical & Regulatory
Implant traceability. Controlled substance law (BtMG). MiGEL reimbursement codes. Oncology drug catalogues. Swiss KVG billing obligations.

No hospital consulting engagement. No six-month discovery phase.
AI brought the SME knowledge. The framework made it actionable.

New domain = new conversation.
Same framework.

jinflow
engine
+
AI SME
domain knowledge
Domain Pack
contracts + probes + i18n
Running
Instance

To enter retail: ask AI about SAP Retail schemas, shrinkage patterns,
planogram compliance, margin erosion — generate the domain pack.

To enter pharma: ask AI about batch records, LIMS formats,
yield calculations, GMP compliance — generate the domain pack.

The framework stays. The expertise is injected.

Each domain pack makes AI smarter
for the next one.

  • Hospital deployment reveals that balance probes need interpretation rules — DRG inpatient vs outpatient. The engine gains the interpretation layer.
  • Retail deployment reveals that shrinkage probes need seasonal adjustment. The DSL gains a seasonal modifier. All domains benefit.
  • Each domain's calibrated thresholds inform the next. “What does normal look like?” gets answered by data, not guesswork.
  • AI accumulates cross-domain patterns. The framework gets better at every domain because it's been to others.

Where jinflow Fits

Six industries. Same engine. Different domain packs.

Material/resource flows + financial reconciliation

Manufacturing
Work orders, BOMs, inventory, shipments
Phantom consumption, BOM vs actual variance, scrap anomalies
Pharma & Life Sciences
Batches, formulations, QC results, distribution
Yield deviation, expiry risk, recall traceability gaps
Facility Management
Service orders, assets, maintenance, procurement
Unused contracts, maintenance backlog trends, cost overruns
Retail & Warehouse
Purchase orders, stock movements, sales, returns
Shrinkage detection, dead stock, supplier delivery anomalies
Public Sector Procurement
Tenders, contracts, invoices, deliveries
Duplicate invoicing, contract compliance, budget overruns
Construction
Projects, material requisitions, deliveries, invoices
Material waste, subcontractor billing anomalies

Every domain has the same shape

Multiple ERPs
different column names
Canonical Model
validated entities
Diagnostic Probes
reconciliation + anomalies
Hypotheses
business language

Multiple source systems → canonical model → financial/operational reconciliation
→ diagnostic probes → management-level hypotheses.

The strongest fit is anywhere you see this shape.
The entity names change. The architecture doesn't.

Same probe types. Different vocabulary.

Probe Type Hospital (nuMetrix) Retail (jinflow)
balanceUsage value vs billed amountReceived qty vs (sold + on hand)
mandatory_itemHip replacement → implantStore type → core assortment
trendMaterial cost drift per monthMargin erosion over rolling window
distribution_outlierDRG cost outlier per caseReturn rate anomaly per product
assessmentCase financial integrityStore health score

The probe compiler generates identical SQL structures.
Only the entity names and thresholds change.

What Cannot Be Copied

The engine is reproducible. The domain packs are not.

Each domain pack encodes
knowledge that accumulates.

  • Regulatory knowledge. Swiss KVG. BetmG. MiGEL. GMP. Food safety. Each domain has its legal landscape. The domain pack encodes it.
  • Calibrated thresholds. SZO's I/O coefficient of 0.39 tells you what “normal” looks like at a Swiss regional hospital. You can't guess this. You earn it from data.
  • Interpretation rules. “A 40% billing delta on a DRG inpatient case is expected. On an outpatient case, it's CHF 2,500 at risk.” The difference between noise and signal.
  • Source-system expertise. OPALE's 5-digit mandate prefix. SAP BWART 301 = TRANSFER. Navision's “Negative Adjmt.” = ISSUE. Earned through ingestion, not documentation.

Open engine. Expert domain packs.

jinflow Engine
Pipeline, compilers, Explorer shell.
Open or source-available.
Drives adoption and trust.

The platform. Free to evaluate.
Domain Packs
Probes, interpretation rules,
regulatory context, calibrated thresholds.
Subscription per tenant.

The expertise. The revenue.
Custom Domains
Consulting engagement to build
a domain pack for a new industry.
AI-accelerated. Weeks, not months.

The growth engine.

The framework
that learns every domain
it enters.

nuMetrix proved it works for hospitals.
The engine is domain-agnostic. The domain packs are pluggable.
AI brings the expertise. Each deployment makes the next one faster.

One framework. Many domains. Infinite expertise.

jinflow.io