Pollenix
III. pollenix.health

AI-native EMR · FHIR R4 · regulated

The EMR that ships with its reasons.

Pollenix MR is a deployable AI-native EMR on FHIR R4. Every clinical surface carries citations — audio spans on scribe notes, FHIR resource ids on prior-auth packets, SOAP text spans on suggested codes.

Built for outpatient primary care, specialty clinics, and inpatient. Audio-span citations on every scribed section. Per-call AuthContext + Why trace on every write. Files stay in your cloud. Pricing is coverage-based, not consumption-based — so the always-on ambient scribe + continuous reconciliation surfaces don't create metering anxiety.

01. The clinical loop

Built for the way clinicians actually work.

Six tightly-coupled surfaces. Each cites the others. Each writes to FHIR. None of the screens require AI to be useful, but every screen is better because AI augments it without taking the pen.

  • I. scribe

    Ambient scribe

    • getUserMedia → MediaRecorder → 16 kHz WAV → STT verbose_json with word-level timestamps.
    • SOAP draft with character-span citations into the transcript. Hover §1 → highlight supporting words.
    • Apply to encounter buffer in one click; post as FHIR Composition with section.extension carrying (transcript_id, span) — portable evidence outside Pollenix's UI.
    pollenix.health · scribe
  • II. problems

    Problem list & reconciliation

    • Decay signal flags active conditions untouched > 180 days. Direct attack on problem-list rot.
    • Scribe-proposed deltas (add / resolve / merge) with one-tap accept. Each delta cites the audio span.
    • Trajectory data layer: onset → recorded → abatement. Encounter linkage in the next sprint.
    pollenix.health · problems
  • III. meds

    Medication reconciliation

    • Continuous diff between MedicationRequest (prescribed) and MedicationStatement (patient-reported).
    • Three columns: reported-only / prescribed-only / reconciled. One-tap promote or stop.
    • Surescripts NewRx mTLS for non-controlled prescribing; EPCS-gated for schedules II–V.
    pollenix.health · meds
  • IV. labs

    Lab inbox + AI trend summary

    • Abnormal-first triage from HL7 interpretation flags + reference-range fallback.
    • AI trend narrative cites Observation ids: 'HbA1c rising from 7.1 to 8.4 over six months [obs-id-1, obs-id-2]'.
    • Close-loop transaction: acknowledge + follow-up ServiceRequest + patient Communication letter under one Why trace.
    pollenix.health · labs
  • V. plan

    Care plan & chronic care

    • CarePlan + Goal with LOINC-coded numeric targets and due dates.
    • Auto-flip on matching lab result: when a new HbA1c lands ≤ target, the goal flips to achieved automatically.
    • Best-effort reconcile triggers on every clinician lab acknowledge. Chronic-care closes on normal review.
    pollenix.health · plan
  • VI. safety

    Allergies & interaction check

    • AllergyIntolerance writes with criticality + severity + reaction.
    • Drug-allergy + drug-drug interaction check at every MedicationRequest authoring.
    • Real licensed DDI (FDB / Lexicomp) ships in a follow-on for Joint Commission certification.
    pollenix.health · safety
02. AI moats

Citations are a FHIR primitive, not a UX flourish.

The audio span the clinician dictated. The FHIR resource id the model read. The SOAP text span that supports each suggested code. Each lives on a FHIR resource extension — verifiable outside our UI, by any downstream consumer.

Chart Review Agent

30-second summary of the patient before the visit. Active problems, medication concerns, abnormal trends, open loops — each citing the FHIR resource ids the model drew from.

Prior-Auth Drafter

Drafts a PA packet for any MedicationRequest or ServiceRequest. Each evidence claim cites a resource_ref. Completeness score + missing-fields list so the biller knows what to chase before submission.

Denial Scrubber

Pre-submit denial probability + gap list against payer-rule heuristics (medical necessity, frequency limits, modifiers, PA on file, duplicates). Each gap shows field / issue / fix / severity.

Code Suggester at sign-off

E/M level + ICD-10 dx + CPT procedure codes from the SOAP body + active problems. Each code rationale links back to the supporting SOAP text span — pairs with audio-span citations for the full audio → SOAP → code chain.

03. Wires in, wires out

Every healthcare integration that matters, real not stubbed.

FHIR R4 bulk export to migrate a chart out of Epic / Cerner / Athena. HL7 v2 MLLP to receive lab results from the analyzer. X12 270/271 to check eligibility. X12 837P/835 to bill. Surescripts for prescriptions. DirectTrust for secure clinician-to-clinician email. All wired against real spec — not vaporware.

FHIR R4

US Core profiles + Bulk Export ($export) for Epic / Cerner / Athena migration.

HL7 v2 MLLP

Inbound ORU^R01 (lab) + ADT^A01/A04/A08 (admit / register / update).

Surescripts NewRx

mTLS prescription send. EPCS-gated for controlled substances.

X12 270/271

Eligibility check before service. Real-time round-trip with the clearinghouse.

X12 837P/835

Claim encode + remittance parse. Invoice flips automatically on paid response.

DirectTrust

Multipart MIME + detached PKCS#7 S/MIME for clinician-to-clinician secure email.

WADO-RS DICOM

Imaging study list + frame fetch via rendered/jpeg endpoint.

Twilio Video

Telehealth room create + JWT access token for the FE Video SDK.

Stripe Connect

Per-tenant Connect account onboarding + webhook signature verification + Invoice flip on payment.

04. Audit & compliance

Designed around the audit, not bolted onto it.

Every clinical surface writes through one audited path. The per-call AuthContext + Why trace architecture isn't a marketing line — it's what makes ONC integrity criteria + EPCS DEA audits tractable in the first place.

  • Per-call AuthContext capability tokens (≤ 30 s TTL, single-use, audit-logged).

  • Why trace on every write — purpose_code (TREAT / BILLING / PAYMENT), subject_id, parent_trace_id chained across multi-step transactions.

  • Separate KMS key + separate audit table from non-Health Pollenix surfaces.

  • BAA available. Files stay in the customer's cloud — Pollenix is deployable, not SaaS.

  • ONC §170.315 criteria mapped surface-by-surface; cert in flight with an ACB.

Pollenix MR

Deploy where your data lives.

Pollenix is sold through AWS and Azure Marketplaces — single SKU, tenant-scoped deploy in your account. Files stay in your cloud. BAA on signature. Coverage-based pricing.