A patient sees a cardiologist across town, lands in the emergency department two weeks later, then comes back to primary care with a medication list that no longer matches the chart. The PCP has 15 minutes. Somewhere in the city, the information exists. In the exam room, it might as well be buried under concrete.
This is the everyday interoperability problem in American healthcare. Medical data exists in overwhelming quantities. We generate labs, discharge summaries, CT reports, operative notes, medication histories, billing codes, portal messages, refill requests, and scanned PDFs until a single patient can accumulate thousands of pages of documentation.
The problem is that the data often moves awkwardly and arrives in a form that rarely fits the clinical moment.
The Records Live in Different Clinical Systems
American medical records live across a patchwork of ambulatory charts, hospital platforms, lab systems, imaging systems, payer systems, health information exchanges, and patient portals. Most of those systems were built to solve real operational problems inside their own setting. A hospital discharge workflow, a primary care medication reconciliation screen, and a specialty practice template all carry different assumptions about what matters.
Patients experience this as friction. Doctors experience it as chart blindness. You know the patient was hospitalized. You know a specialist changed the dose. You know the CT report matters. Then you spend the visit hunting through a portal, calling for records, or clicking through an outside document viewer that was never designed for a rushed follow-up visit.
Even when two organizations use the same EMR, the exchange can be thinner than people expect. Different builds, different versions, different local templates, and different custom fields change the shape of the data. The words may arrive. The meaning still has to be reconstructed.
The Law Is Finally Pushing Records Outward
The federal government has been moving, with real force, toward a healthcare system where records can follow the patient. The ONC Cures Act Final Rule was designed to give patients and clinicians secure access to electronic health information and to address information blocking. ONC describes information blocking as practices that are likely to interfere with access, exchange, or use of electronic health information, unless an exception applies.
TEFCA is trying to create a national floor for trusted health information exchange so that data can move more consistently across organizations. FHIR APIs are becoming part of that exchange. CMS has also pushed payer and provider access APIs through its Interoperability and Prior Authorization Final Rule.
This matters. It is easy to be cynical about healthcare technology policy, and clinicians have earned that cynicism. Still, the direction is clear: records are supposed to be easier to retrieve, easier for patients to access, and easier for clinicians to use across care settings.
See What Grail Does With the Clinical Conversation
Grail turns the visit into structured clinical documentation, patient instructions, and follow-up work that fits the way physicians actually practice.
Explore FeaturesRetrieving the Record Is Only the First Step
A CCD dump is better than nothing. It is also a brutal thing to hand to a doctor who is already behind. I have seen outside records arrive as a long trail of duplicated problem lists, medication entries with no clean source of truth, hospital notes copied forward for days, and PDFs that bury the useful sentence on page 17.
The receiving system can store that material. The physician still has to make sense of it. Did the patient actually have heart failure, or did the diagnosis appear because someone ordered a BNP during a pneumonia admission? Was metoprolol stopped because of bradycardia, fatigue, hypotension, or because the discharge med reconciliation got messy? Which colonoscopy report is the current one?
Interoperability solved at the transport layer still leaves the clinical layer unsolved. HL7 and FHIR can define resources and fields. By themselves, they cannot produce a usable story.
AI Is Built for the Second Half of the Problem
AI is very good at the work that makes interoperability clinically useful: reading large volumes of messy text, extracting the relevant facts, comparing conflicting versions, and reformatting information into a different shape. That is exactly what healthcare needs after the outside records arrive.
Instead of 1,200 pages from the hospitalization, the primary care doctor needs the reason for admission, the key findings, medication changes, pending tests, follow-up needs, and a short explanation of what changed from the last known baseline. The endocrinologist needs the diabetes trajectory, medication intolerance history, renal function, hypoglycemia pattern, and the reason the GLP-1 was stopped, rather than every outpatient note from the last five years.
AI healthcare interoperability means the receiving system can ingest the pile, understand the clinical signal, and reshape it into a concise summary that fits the visit. That summary can be reviewed in a few minutes. The doctor can ask better questions. The patient stops serving as the interface between two incompatible software systems.
The Data Shape Problem Is Where AI Becomes Practical
Health systems talk about interoperability as if it were mainly a pipe problem. Connect system A to system B, send standardized resources, and the work is done. Anyone who has built clinical software knows better.
The sender's chart and the receiver's chart rarely think the same way. One workflow treats obesity medicine as a specialty template. Another buries the relevant medication failures in free text. One chart has a clean surgical history field. Another stores the bariatric operation in a scanned operative note, a problem list item, and a billing code. The receiving EMR may accept the data, then place it somewhere that is technically available but clinically easy to miss.
AI can translate between those clinical shapes. It can turn a discharge summary into a primary care follow-up brief, a specialist consult into a problem-specific history, or a chaotic medication trail into a reconciliation task. That is more valuable than another tab.
What This Changes in a 15-Minute Visit
Imagine the patient in front of you was hospitalized across town two weeks ago. Today, the chart opens with a clean outside-care summary: acute kidney injury during admission, lisinopril held, creatinine improved from 2.1 to 1.3 before discharge, repeat BMP recommended in one week, CT abdomen showed an incidental 1.8 cm adrenal nodule, no endocrine workup ordered.
That changes the visit immediately. You review the blood pressure, reorder the BMP, decide whether to restart the ACE inhibitor, explain the adrenal finding, and schedule the right follow-up. Without the summary, the visit becomes detective work. With the summary, it becomes medicine.
This is where AI can make interoperability feel real to clinicians. The value comes after the data transfer, when the right part of the record shows up at the moment of care.
The Medical Record Should Follow the Patient
The future version of this system looks obvious once you see it. A patient's medical record follows them wherever they go. When they change clinics, see a specialist, visit an urgent care, or land in a hospital across town, the receiving clinician gets a digestible summary of the relevant history, recent care, active medications, open loops, and clinical uncertainty.
That future will still need privacy rules, consent, audit trails, patient control, and serious safeguards. Moving health data around carelessly would be a disaster. Leaving clinicians to rebuild the story from scattered records keeps the burden in the exam room.
AI will not magically fix American healthcare interoperability. It gives us the missing layer between record retrieval and clinical understanding. That layer is the difference between handing a physician a thousand pages and handing them the five minutes of context they needed before they walked into the room.
Think about that the next time you see a doctor. If that physician could see a concise, accurate summary of the care you received elsewhere - the specialist visits, the hospital admission, the medication changes, the tests still pending - the visit would start in a different place. I still do not know how fast that future will arrive. I know clinicians and patients need it now.