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 Ingest & mapping QA Data Store
Integration
Source & target EMRs Source patient import Bulk import manager Streaming data Bulk export manager Streaming export
System
Control panel Model monitoring
Governance
Registry & preflight Fairness screen Ambient consent gate Source attributes EMR conformance Retention Audit
Documentation
System documentation Implementation & setup guide Feature documentation Model documentation System architecture EMR connections Emulators & non-PHI setup Compliance & responsibility Attribution
Ryan GomezRole
Signed in — Ryan Gomez · System Administrator
You stay on this screen — each role sees it as their own permissions allow. The assistant thread and the patient in context start fresh.
Working archetypes — combinations people really hold
One role at a time — each role on its own
DocumentationDocs
System documentation Implementation & setup guide Feature documentation Model documentation System architecture EMR connections Emulators & non-PHI setup Compliance & responsibility Attribution
Repository

Streaming data manager

Real-time feeds beside the batch pipeline — admissions and discharges, lab results, claims remittances. Streaming has no clean-run boundary, so each partition keeps its own checkpoint with explicit gap accounting, never sharing the bulk pipeline's watermark. A gap is surfaced loudly rather than interpolated.

10,407
events today across all feeds
6
partitions under checkpoint discipline
1
open gaps — surfaced, not interpolated

Partitions

FeedPartitionCheckpoint Latest offsetLag Events todayLast eventState
ADT admissions/discharges 0 8,842,310 8,842,310 0 1,204 13:58 healthy
ADT admissions/discharges 1 8,790,522 8,790,522 0 1,187 13:58 healthy
ADT admissions/discharges 2 8,811,948 8,811,990 42 1,163 13:51 gap — attention
Offsets 8811949-8811990 unacknowledged after broker restart. Surfaced loudly, never interpolated - an interpolated gap is a silent data-integrity failure that looks like a healthy pipeline.
Claims 835 remittance 0 902,114 902,114 0 214 12:40 healthy
ORU lab results 0 4,410,221 4,410,221 0 3,388 13:59 healthy
ORU lab results 1 4,398,710 4,398,710 0 3,251 13:59 healthy
Why two disciplines. The bulk pipeline has a clean-run boundary and a watermark; a stream has neither — only per-partition offsets and the honesty to account for every one. The gapped partition above holds its checkpoint until the missing offsets are recovered or the loss is formally accepted and recorded; it does not skip ahead.

Source EMR lookup — pull one patient now

Between bulk runs, new patients exist at the source that PHI AI has not seen. Search the source EMR directly and import an individual patient over the per-patient read path — the same path bulk-less vendors force for entire populations.

An individual import runs read → encrypt → store → index for one patient and lands in your session's sandbox: the patient appears in Patients & charts, the import is audited (ingest.patient_imported), and the next reconciliation on the Bulk import manager counts them.