Not based on a specific customer deployment. The company, figures, decisions, and outcomes below are a constructed example that shows how Intended’s mechanism applies to this kind of problem. They are not measured results from a named customer. Where we publish a real, attributed customer outcome, we will say so explicitly.
HIPAA-Aware Governance for AI Patient-Data Operations
Consider a health system whose AI agents handle appointment scheduling, medical record access, billing automation, and clinical decision support — all touching PHI under HIPAA.
01 · the challenge
The kind of problem this addresses.
02 · how it works
See the difference.
AI agent requests patient records
Patient: J. Doe, MRN: 847291
System returns records
No authorization check, no logging
AI processes PHI
Full record access, no scope limits
Flat log entry (maybe)
No evidence chain, no tamper detection
03 · the solution
What they deployed.
- — Started from Intended's healthcare policy pack (deny and escalate rules for PHI export and access) and added the health system's own rules
- — Full-record PHI access is CRITICAL risk and needs a clinician's approval
- — Each approval yields a single-use authority token for one patient, valid for at most five minutes
- — Agents call Intended through the SDK before reading from Epic or the scheduling system. Intended never touches the EHR; the health system's own EHR integration service verifies the token before it returns records
- — Requests send record identifiers and the access purpose to Intended, not clinical content
- — Every PHI access decision is written to a tamper-evident, hash-chained audit trail and can be exported as an evidence bundle
04 · implementation
From zero to governed.
Phase 1
Assessment
Map AI agent PHI access patterns. Identify unprotected access paths across clinical systems.
Phase 2
Deploy
Install the healthcare policy pack. Configure PHI access policies, clinician approval routing, and token checks in the EHR integration service.
Phase 3
Validate
Run policy simulation on recorded access requests to check outcomes before enforcing.
Phase 4
Enforce
Switch to enforcement. Every PHI access then carries a decision record in the tamper-evident audit chain.
05 · illustrative outcomes
What this is designed to deliver.
Modeled figures for this scenario — what the workflow above is built to achieve, not measured results from a named customer.
0%
PHI access token-gated
Clinician approval for CRITICAL access
0%
PHI access logged
With tamper-evident evidence
0 min
Max token lifetime
Single-use, no lingering access
0%
Decisions with a receipt
Verifiable against the tenant's public keys
06 · decision replay
Example decisions, full trace.
Sample decision records that show the shape of the evidence Intended produces. Illustrative, not drawn from a live customer’s logs.
healthcare.phi.record-accessRISK: 72/100ALLOWAI scheduling agent requests patient demographics for appointment confirmation
Resolved by: Policy: demographics-only access auto-approved with single auth
healthcare.phi.clinical-recordsRISK: 95/100ESCALATEAI clinical support agent requests full medical history for treatment recommendation
Resolved by: Dr. Martinez (clinician approval)
healthcare.phi.billing-dataRISK: 58/100ALLOWAI billing agent requests insurance information for claims processing
Resolved by: Policy: billing scope access with system auth
healthcare.phi.bulk-exportRISK: 99/100DENYAI analytics agent requests bulk patient data export for population health report
Resolved by: Policy: bulk PHI export requires IRB approval + CISO sign-off
the takeaway
The hard part of HIPAA for AI is independent verifiability: an auditor should be able to confirm every AI access to patient data without trusting self-reported logs. Signed decision receipts and a tamper-evident hash chain make each access provable rather than asserted.
Why this pattern matters — not a customer quote.
Start protecting patient data
Free to start. See every AI decision from day one.