Skip to main content
Health benefits are regulated. This page summarizes the plan structure, data handling boundaries, and go-live gates that affect the integration.

Who is what: the plan structure

Use these roles when mapping the plan into regulated surfaces:
RoleWhoWhat it means
Plan sponsorThe employerThe employer sponsors the plan and owns the final plan decisions.
Benefits operationsPrescienceWe support rating, enrollment, claims workflows, care navigation, funding operations, and compliance coordination.
Risk protectionLicensed carrier partnerCarrier-backed protection helps cap downside exposure for eligible high-cost claims.
Use the approved plan display names from GET /plans in customer-facing surfaces. The planType value is backend classification for eligibility, quoting, and regulated workflows; do not expand it into marketing copy.

Quotes are indicative

Every quote carries "pricingBasis": "indicative" and ships with its own assumptions strings. The contract:
  • Quote numbers are indicative pending underwriting confirmation.
  • They become binding only at enrollment plus underwriting sign-off.
  • Material census changes re-rate.
Keep that framing attached to any displayed quote amount.

Data handling

What you send us

Census data (names, dates of birth, ZIP codes, emails, dependent relationships) is enrollment and eligibility data. Before live mode is enabled, Prescience and your company execute:
  • a Business Associate Agreement (BAA), and
  • a data processing agreement
covering exactly this data. This is why psk_live_ keys return 403 live_mode_disabled until the partnership paperwork is done: the gate is contractual, not technical. Test mode is for development against the full API without that data relationship in place; use realistic-but-fake people in test mode if your policies require it.

What we send you

The other direction is restricted by design:
  • Aggregate and de-identified only. GET /account returns funding, premium, spend, and savings aggregates: never named claims, never diagnoses, never any individual’s health data.
  • k-anonymity suppression. When a spend category could be traced to a small number of members, the API suppresses the categories (spend.suppressed: true) instead of returning them. You don’t implement this; your only job is to render the suppressed state.
  • The roster is administrative. GET /members carries enrollment status and a dependent count: no dependent identities, no health information.
The practical consequence: your platform never holds PHI from Prescience, which keeps your own compliance surface where it already is (HR data) rather than expanding it into claims data.

Go-live checklist

  1. Core flow green in test mode (group → census → quote → enrollment), webhook signatures verified end to end.
  2. BAA + data processing agreement executed.
  3. Required commercial and operational agreements completed.
  4. Prescience enables live mode and issues your psk_live_ key.
  5. First real employer group created with the live key.
Items 2–4 can run in parallel with implementation.
Questions about a specific employer situation, state, or filing? Ask your Prescience partner engineer. Nothing on this page is legal advice for your own obligations as an HR platform.