φ(ai) PHI AI
Menu
Clinical workspace
Patients & charts PHI AI assistant Signature queue Ambient documentation Imaging worklist Summarization Inbox triage Patient instructions
Revenue & records
Claims & billing Prior auth & appeals Documentation gaps Release of information Sensitive categories
Population
Cohort builder No-show risk Scheduling optimization Trial pre-screening Quality measures
Data Orchestration
Orchestration Connected systems Patient link set Practitioner crosswalk Permission lattice What each system holds Run ledger Ingest & mapping QA Data Store
System
Control panel Model monitoring Components
Governance
Registry & preflight Fairness screen Ambient consent gate Source attributes Retention Audit
Documentation
System documentation Implementation & setup guide Feature documentation Model documentation System architecture EMR connections Orchestration & identity Emulators & non-PHI setup Compliance & responsibility Attribution

Control panel

The System Administrator's seat. Everything on the platform can be controlled from here — the live model, the outbound feeds, the import hold, the integrity checks — and none of it quietly: full control is not exemption, and every switch thrown on this page lands on the audit trail under your name.

WARNING
Command deck — every capability slot, and the controls that decide what serves it
14slots
0running
11idle
0no model
0out of service
3open alerts
0inferences

One card per capability slot. Binding chooses which registered model serves the slot; power takes the capability out of service without touching the model behind it. Both are audited under your name the moment you commit them, and every screen that uses a slot tells its reader when the slot is not serving — nothing here fails quietly. The readings are the console's own, so this page and Model monitoring can never disagree.

Denial risk
claims · predictive
ALERTING
Denial risk model v1.0
built-in default for this slot · built-in
0.637 Discrimination (AUC)
0 inv never run 1 alert →
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Sensitivity classification
segmentation · classifier
ALERTING
Sensitivity classifier v1.4
built-in default for this slot · built-in
enforced Fail-closed posture
0 inv never run 1 alert →
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Coding integrity
coding · classifier
ALERTING
Coding integrity model v1.0
built-in default for this slot · built-in
0% Findings answered
0 inv never run 1 alert →
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
No-show risk
noshow · predictive
IDLE
No-show risk model v2.1
built-in default for this slot · built-in
42% Mean predicted risk
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
PHI AI assistant (retrieval)
assistant · retrieval + LLM
IDLE
Claude Sonnet 5 v1
built-in default for this slot · Anthropic API
every claim Attribution gate
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Release-of-information validation
roi · classifier
IDLE
ROI requirements validator v1.0
built-in default for this slot · built-in
none Approval authority
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Ambient listener
ambient · speech recognition
IDLE
Ambient listener — faster-whisper v1.2
built-in default for this slot · Self-hosted (MIT)
no captures yet Consent before capture
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Imaging read assistant
imaging · vision
IDLE
Imaging read assistant v1.0
bound by an operator · Anthropic (your BAA)
none Diagnostic authority
0 inv never run
1 alternate registered and enabled
Instrument → How it works Configure
Patient instructions
instructions · generative + checker
IDLE
Patient instructions writer v1.0
bound by an operator · organization-managed
67% Checker pass rate
0 inv never run
1 alternate registered and enabled
Instrument → How it works Configure
Prior-auth evidence assembly
priorauth · retrieval
IDLE
Prior auth evidence assembler v1.0
built-in default for this slot · built-in
no drafts yet Clinician acceptance
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Inbox triage
triage · deterministic ranking
IDLE
Inbox triage ranker v1.0
built-in default for this slot · Built-in
|out| = |in| Set preserved
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Scheduling optimization
scheduling · optimization
IDLE
Scheduling optimizer v1.2
built-in default for this slot · built-in
read-only Calendar authority
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Terminology mapping
ingest · mapper
IDLE
Terminology mapper v3.0
built-in default for this slot · built-in
100.0% Coded coverage
0 inv never run
No alternates enabled — register one in the Model registry
Instrument → How it works Configure
Ingest & mapping QA
ingestqa · rules + LLM reviewer
IDLE
Ingest QA reviewer v1.0
bound by an operator · organization-managed
9 of 10 Rules clean
0 inv never run
1 alternate registered and enabled
Instrument → How it works Configure
What binding does not do. Putting a model into a slot is an operational decision, not a clinical one. It does not validate the model on your data, it does not run the fairness screen, and it does not discharge the oversight duty that sits with the organization. A predictive model that has not been through Registry & preflight and the fairness screen should not be serving patients, whatever this panel lets you bind.