Care moves. Software rarely follows.
Care crosses buildings, organizations and professions in a single week. The systems that record it were designed one building at a time — and that mismatch, not a shortage of software, is the problem worth solving.
Consider an ordinary week in the life of one patient. An older adult, recently discharged after a hospital stay, living at home with a chronic condition and a family that worries.
On Monday a caregiver arrives at the front door, helps with medication, notices that the ankles are more swollen than last week, and writes it into the home care agency's visit record. On Tuesday a phlebotomist collects a sample that becomes a result in a laboratory system and, three days later, a PDF. On Wednesday a wearable logs a fourth consecutive night of broken sleep and a resting heart rate drifting upward; the data lives in a consumer app that nobody clinical has ever opened. On Thursday the patient attends a clinic appointment that was booked six weeks earlier for an unrelated reason. On Friday a fifteen-minute teleconsultation takes place with a physician who has seen none of the above and therefore asks the only question available: how have you been getting on?
The patient answers from memory.
Memory is now the integration layer.
Nothing in that week was a failure of care. Every professional involved did competent work and recorded it properly in the system they were given. The caregiver's note is in the caregiver's system. The result is in the laboratory's system. The wearable data is in the wearable's app. The clinic's record contains the clinic's version of the patient, which is the only version the physician on Friday can see. Six honest records, none of which is the record of the week.
Every hop is a boundary, and context dies at boundaries
A system boundary is not a technical inconvenience. It is a place where meaning is deliberately dropped.
What crosses a boundary is almost always the datum: a value, a document, a fax, a discharge summary, an attachment. What does not cross is the context that made the datum mean something — who observed it, what they were looking for, what it looked like last month, what the patient said while it was being taken, what the observer decided to do next and why. The receiving system has no field for most of that, and no reason to trust it if it did.
So each hop performs the same lossy compression. A caregiver's observation that the swelling is worse than last week becomes, downstream, a note that swelling was present — if it travels at all. A borderline laboratory value arrives without the three previous values that would have shown it moving. A school nurse's log of a child's recurring complaint stays inside the school. A wearable's fortnight of declining sleep never becomes clinical context, because there is no path by which it could.
The result is a peculiar kind of poverty. There is more health data about this patient than about any previous generation, and the person best placed to act on Friday afternoon has almost none of it. Not because it is confidential, not because it is unavailable in principle, but because it is spread across six systems whose only common key is a human being's memory of their own week.
Fragmentation is not the absence of technology.
It is the accumulated shape of a great deal of technology, each piece of it built well for one room.
Care moves. Caretech keeps the thread.
Follow one person through one week — hospital, home, rehabilitation, diagnostic center, follow-up. Five rooms, four organizations, one record accumulating the whole way. The setting changes. The thread does not break.
Hospital — the thread begins
A discharge that used to be a fax is an encounter on a record. Vitals at discharge, the changed medications, the plan for home — entered once, attached to the person rather than the building. The thread begins before the wheelchair reaches the door.
Home — the reading and the note join
A connected cuff on the kitchen table sends the morning reading into the same series — not into a consumer app nobody clinical opens. The caregiver at the front door writes the swollen-ankles note into the same record. Two boundaries crossed; nothing dropped.
Rehabilitation — progress joins, a pattern is noticed
Each session's note lands on the thread beside the readings it explains. Quietly, the platform recognizes context across settings — this week's readings drifting against the discharge baseline — and marks it for review. It surfaces; it does not decide.
Diagnostic center — the result lands in the series
A sample becomes a result that arrives already in the record, trended against this patient's own baseline — not a PDF in a portal three days later. The laboratory kept its rigor; the person kept the result.
Follow-up — a clinician reviews the week, and acts
At the follow-up visit the clinician sees the whole thread — discharge, readings, notes, result, the flagged pattern — and makes the call a qualified person is there to make. The decision returns to the record, where the next setting will find it. The thread holds.
01 · Hospital
The thread begins
A discharge that used to be a fax is an encounter on a record. Vitals, changed medications, the plan for home — entered once, attached to the person rather than the building.
On the thread encounter → record
02 · Home
The reading and the note join
A connected cuff on the kitchen table sends the morning reading into the same series. The caregiver at the front door writes the swollen-ankles note into the same record.
On the thread encounter + device reading + caregiver note
03 · Rehabilitation
Progress joins, a pattern is noticed
Each session's note lands on the thread beside the readings it explains. Quietly, the platform recognizes context across settings and marks it for review. It surfaces; it does not decide.
On the thread + session note pattern marked for review
04 · Diagnostic center
The result lands in the series
A sample becomes a result that arrives already in the record, trended against this patient's own baseline — not a PDF in a portal three days later.
On the thread + lab result
05 · Follow-up
A clinician reviews the week, and acts
The clinician sees the whole thread — discharge, readings, notes, result, the flagged pattern — and makes the call a qualified person is there to make. The decision returns to the record. The thread holds.
On the thread reviewed by a clinician action → record
Every entry above crossed a building, an organization or a profession — and stayed on one record. That thread is the product. The usual way to chase it is to buy a tool for every gap. Watch what that does instead.
The instinct to buy another point solution
The usual response to a boundary is to buy something that spans it. Watch what that does to the number of boundaries.
A tool for every gap
A remote monitoring product for the wearable gap. A referral tool for the clinic gap. A coordination layer above the layers, and a messaging application so the people running the other six can talk. Each purchase is defensible on its own terms.
Each tool adds boundaries
Every new system brings its own identities, its own permissions, its own notion of what a patient is — and a treaty with each of the others. Integration projects do not remove boundaries; they negotiate treaties across them, and the number of negotiations grows with the square of the number of tools. Seven tools is twenty-one treaties, each one holding only while both sides hold still.
Hold four things in common
One identity, one record, one permission model, one trace. Held in common, the tools stop being systems and become workflows on one substrate — and the treaties disappear with the boundaries they crossed.
A new setting is one edge
Adding a school health service or a correctional program is an extension, not an integration: a workflow, a set of roles, a set of rules. One edge to the record — not seven new negotiations.
This is why "we have integrations" and "we have a platform" are different claims. Interoperability standards — HL7, FHIR, exchange with electronic health record, laboratory and imaging systems — are necessary and worth building properly: they are how a platform reaches the world it did not build. They are not a substitute for having something coherent on the inside for the world to reach.
What has to be held in common
A platform, in the sense worth defending, is not a collection of modules from one vendor. It is a small number of things held in common by every workflow that runs on it. Four of them do most of the work.
Workflows
Telehealth, home care, scheduling, screening — every workflow stands on the same four commons.
One identity
A person appears once, and is recognized in every workflow.
One record
The series in one place, context intact.
One permission model
Role and relationship, enforced in the database.
One trace
Who did what, when, and on what basis.
A new setting
Slides in as an extension standing on the same four pillars — not an integration with the other workflows.
One identity
A person appears once — patient, clinician, caregiver, family member with permission to see a subset. Duplicated identity is a permanent clerical tax, and the duplicate record is always the one holding the fact you needed.
One record that accumulates
The kitchen-table reading, the front-door visit, the school nurse's log and the laboratory result land in the same longitudinal record, keeping the context and not just the value. A trend only exists if the series is in one place.
One permission model
Access is a property of role and relationship, enforced at the boundary of every request and in the database itself — widening what the system knows does not widen who can read it.
One trace
Who did what, when, and on what basis, recorded as a first-class part of the workflow. A system that cannot account for its own activity has replaced six defensible silos with one indefensible pool.
The argument, stated plainly
The founding argument of this company is not that healthcare software is bad. Much of it is very good at what it was scoped to do. The argument is that the scope was wrong: software was organized around institutions, and care is organized around people, and people move.
A patient's week is a single continuous thing that happens to be distributed across a home, a clinic, a laboratory, a school and a facility. The record of it should be a single continuous thing too — carrying the patient, the professional, the device and the decision on one system, so that what happens in a home is visible in a clinic, what a nurse documents at a school is available to the physician who follows, and what the data shows afterwards reaches the qualified person who is allowed to act on it.
That is a harder engineering problem than another point solution, and it is the only version of the problem whose solution does not create the next boundary.
Care moves. That is not a defect to be managed at the seams. It is the design brief.
Your care moves. Watch the record follow.
One record across settings is an argument until it runs in your buildings. Tell us where a patient's week actually goes — home, clinic, laboratory, school — and we will show you the record that follows it, and the point solutions it retires.