Every fact carries its source. Every source is checked against what it's actually licensed for. PHI is handled in memory under a signed BAA, every access is audited, and every recommendation is built to end with a clinician's sign-off.
These aren't aspirations bolted on at the end — they're how the platform is built to work.
Protected health information is processed in memory and never written to a persistent member cache.
Every access to patient data is logged, so there's always an answer to "who saw what, when."
ContextRx powers your agent via MCP — you build, host, and run it. We never own, operate, or run the agent, or make the clinical call.
No PHI-touching call happens without a consent check first. It's an invariant of the platform, not a setting someone can forget to turn on.
Every PHI-touching adapter verifies consent before it makes the call — no exceptions.
The merger runs before enrichment, so enrichment never sees raw provider records.
The resolved view is used in the moment; only PHI-stripped output is ever cached.
Every call is written to the audit trail, scoped to the grant that authorized it.
"Free to read" rarely means "free to feed an AI." Every source is tagged with what it's actually licensed for — and anything an AI can't legally use simply isn't in the platform.
Public, redistributable sources, served directly by the platform.
Served directlyUsable only within non-commercial terms, and scoped accordingly.
Scope-limitedFlows through your own license as a BYO pass-through, never resold.
BYO licenseSources whose terms forbid AI use are excluded outright — never ingested.
ExcludedContextRx is HIPAA-aligned and signs a BAA on every live account. We'd rather under-claim than overstate where we are.
Want the details — the BAA, the consent model, the governance registry? Talk to our team and we'll walk you through it.