Healthcare · HIPAA · Interoperability
Where the record is somebody's life
Healthcare technology is the case where every constraint is real at once: the data is regulated, the integration is decades deep, and the failure mode is a person. This writing comes from modernising a $3.1B multi-brand health plan and building AI that keeps PHI on the premises, not from reading the standards.
In healthcare the integration is never the hard part — agreeing what the record means is.
Two systems can exchange a perfectly valid message and still disagree about who the patient is, which encounter this was, and whether the result is current. Most of the difficulty in this field is semantic rather than technical, and most of the risk arrives when a system acts confidently on a record it has misread. The writing here is about the joins, the ownership of them, and keeping regulated data where it is supposed to be.
The frame
Four things that decide a healthcare build
Read the clinical record piece →Who the patient is
Identity resolution across systems that never agreed on it. Every downstream correctness question depends on this one, and it is rarely solved once.
What the record means
A valid HL7 or FHIR message is not a shared understanding. The semantics are where interoperability projects actually fail.
Where the data may go
HIPAA, residency and the question of what leaves the building — which is why the AI work here is local-first rather than retrofitted to be.
Who owns the join
When two records are merged, somebody is accountable for the rule that merged them. Usually nobody is named, and that is the finding.
Flagships
Keeping the data in, and getting the record right
The Data Never Leaves the Box
AI on regulated data without third-party egress: what local-first actually requires, what it costs, and what it buys in a compliance conversation that would otherwise end the project.
Wiring the Clinical Record
The clinical record as an integration problem and a semantic one: what the standards give you, what they do not, and where the disagreements between systems actually live.
Why it fits
Why this is written from inside
A $3.1B health plan, modernised
Multi-brand payer systems unified, with roughly half the operational efficiency gain coming from the unification itself — the work behind several of these pieces.
Payer and provider, startup and enterprise
Series A–C EHR and RCM companies scaling ten-fold, and enterprises where the audit sets the pace. The right move differs and knowing which is the job.
PHI as a starting constraint
The AI systems here were built where PHI is present, so no third-party egress and a clean residency story are the design premise rather than a later concession.
Standards as a floor, not a finish
HL7, FHIR and TEFCA are where the work starts. What the writing is about is the part after conformance, where the systems still disagree.
Field notes
Healthcare technology writing
See all →- Technology Responsibility: A Compliance and Security StudyData
- NIST and Trustworthy AI: The Lab That Never Fines AnyoneAI Governance & Standards
- FedRAMP and GovRAMP: Evidence as a CommandGovernment Cloud Authorization
- DICOM and PACS on MAX4: Pictures Before PatientsMedical Imaging & De-identification
- Identity Data in Marketing and Fraud: The Web of IdentityIdentity Data & Fraud
- Virtual Cell Models: The Bottleneck Is the DataHealthcare AI & Biology Data
- Reading Healthcare-AI Evidence: One Arm Is Not EvidenceHealthcare AI Evidence
- Wearable Health Labels: Cleared, Certified or WellnessWearable Health Regulation
- AI and Healthcare Payment: The $942 Million Nobody CheckedHealthcare AI & Payment
- Local-First Medicine: The Box in the ClinicLocal-First Clinical AI
- Patient Records and AI Agents: The Patient Holds the KeysPatient-Gated Health Records
- AI Voice Agents in Medical Practices: The 81% ProblemHealthcare Voice AI
- Clinicians and AI: What They Actually AskHealthcare AI Evidence
- AI Agent Liability: Who Goes to Prison When the Agent Breaks In?AI Agent Liability
- MARA Demo Update: Earned Authority and a Person on CallHealthcare AI Design
- MARA, the Medical Agentic Resident Assistant: Four Generations, One LineHealthcare AI Design
- MAX4 and MARA: The Ladder Starts at OffMAX4, Trust & Clinical AI
- MAX: The AI Resident, and the Road to MARALocal-First AI & Healthcare
- MARA: A Medical Agentic Resident Assistant That Learns From Every Encounter and Never Makes the DecisionHealthcare AI Design
- Epic Migration Planning: Twelve Thousand Questions, One DateHealthcare IT Program Planning
- EHR Migration Economics: The Cash Trough Nobody BudgetedHealthcare IT Economics
- Reasoning Without a ReasonerReasoning & Ontologies
- The Data Never Leaves the BoxHealthcare & Finance Compliance
- The Crucial Role of HIPAA in Healthcare Software DevelopmentHealthcare
- The Promise of AI in Healthcare Transformation from data to diagnosisHealthcare
- The Untapped Potential of Software Developers in Improving Patient OutcomesHealthcare
- Unlocking Healthcare Data: The Intersection of HL7, FHIR, and HIPAA ComplianceHealthcare
- Unmasking Healthcare Costs: The Urgent Need for TransparencyHealthcare
- What Startups and Multi-Billion Healthcare Enterprises Can Learn From Each OtherHealthcare
- Wiring the Clinical RecordHealthcare
- Connecting the Record and the ContentData
- From Data Revolution to Data Protection: Safeguarding PII and PHI in TechnologyData
- Who Owns the JoinData
- The Window That Only Opens One WayNarrative
- Engineered Cognition · Healthcare EditionNarrative
- Unveiling the Evolution of Work: Impact, Meaning, and Healthcare TechnologyCareer & Leadership
Engage
Building where PHI and the clinical record are involved?
Interoperability and identity review, a local-first AI design that survives a compliance conversation, or a second opinion on a modernisation already under way.
This site: HTTPS only · no third-party trackers · no analytics cookies · security.txt