Every emergency physician starts a shift with a limited supply of something we like to call “brain juice for the day.”
It is not a medical term, of course. It is the mental energy required to assess a sick patient, recognize subtle clinical deterioration, interpret an ECG, decide whether a patient needs intubation, weigh competing diagnoses, communicate difficult decisions, and make the next clinical decision when several things are happening at once.
In an emergency department, that reserve gets used quickly.
A resuscitation consumes it. A difficult diagnostic problem consumes it. A critically ill patient whose condition is changing by the minute consumes it.
But so do things that are not necessarily clinical decisions.
Finding an ICU bed. Calling multiple departments. Checking whether a troponin has resulted. Searching for an old ECG. Going through printed medical records. Arranging an ambulance for a transfer. Finding a ventilator. Writing documentation because an electronic medical record (EMR) is unavailable.
None of these tasks are unnecessary. They are all part of making sure the patient receives appropriate care.
But they still consume the same limited cognitive reserve.
In my experience, approximately 65% of my working time is directly related to patient care – assessing patients, reviewing investigations, making clinical decisions, treating them, reassessing their response, communicating with patients and families, and coordinating their clinical care. The remaining 35% is spent on work that is necessary for patient care but is not direct clinical decision-making: documentation, finding previous records, checking investigation results, arranging beds and transfers, coordinating logistics, and other administrative tasks.
That means more than one-third of a physician’s working time can potentially be consumed by the infrastructure surrounding patient care rather than patient care itself. And that raises an interesting question:
If technology could take some of that work away, could we preserve more of the physician’s “brain juice” for the moments when it actually matters?
The emergency physician is also an administrator
Emergency medicine is often described as a specialty built around rapid clinical decision-making.
But the reality is that an emergency physician is also constantly coordinating the hospital around the patient.
Consider a patient presenting with necrotizing fasciitis complicated by septic shock and multiorgan dysfunction.
The clinical priorities are immediately clear: the patient needs urgent surgical assessment and source control, intensive care, antimicrobial therapy, and potentially support for multiple failing organs.
But implementing that plan may require multiple conversations.
The emergency physician may need to call surgery, ICU, nephrology, anesthesia, and other relevant teams, explaining the patient’s history, examination findings, investigations, treatment given, and current clinical status to each team separately.
The same information may have to be communicated several times.
This is an important part of the job. But it is also an example of where better technology could reduce duplication without reducing clinical communication.
Imagine that a structured summary of the patient could be securely sent to the relevant teams’ phones or computers.
The surgeon could review the case before calling the ED.
The intensivist could see the relevant investigations and current treatment. The nephrologist could review the renal parameters.
The subsequent phone call would then be about clinical discussion, recommendations, and decision-making, rather than spending the first few minutes repeating the basic history. The conversation remains personal.
Technology simply makes sure everyone starts the conversation with the same information.
Waiting for information is also work
Another surprisingly large amount of time in an emergency department is spent checking whether something has happened.
Has the troponin come back?
What is the potassium level?
Is the hemoglobin available?
Has the CT been reported?
Can we access the MRI findings?
Has the radiologist finalized the ultrasound report?
Has the laboratory released the result?
These checks may seem insignificant individually. But across dozens of patients and an entire shift, they add up.
This is particularly important for investigations where the result directly influences the next clinical decision.
Instead of repeatedly checking the EMR or laboratory system, an integrated system could notify the treating team as soon as a clinically relevant result or report becomes available. The physician should not have to repeatedly search for information that the hospital system already knows has become available.
So what could technology actually change?
The solution isn’t simply “more AI.”
Different problems require different technological solutions.
1. A better, truly integrated electronic medical record
One of the biggest problems in healthcare is that information exists, but is difficult to access. A patient may have been admitted several times previously. Their old ECGs, echocardiograms, CT scans, MRI reports, discharge summaries, laboratory results, and medication history may all exist somewhere.
But “somewhere” isn’t necessarily useful.
If the information is stored in paper files, different hospital systems, different departments, or disconnected software platforms, the physician still has to spend time finding it. A better EMR should bring relevant information together.
A physician seeing a patient should be able to access previous investigations and medical records without searching through stacks of printed documentation or calling another department to locate a report.
Ideally, this information should also be available across hospitals, with appropriate privacy and access controls.
A patient transferred from one hospital to another should not have to carry their entire medical history as a collection of printed reports.
The same integrated system could also improve communication between departments. Rather than calling multiple teams individually to provide the same basic information, a structured patient summary could be made available to the relevant teams. The physician can then speak directly with the appropriate specialist to discuss the clinical situation and obtain recommendations.
Technology should not replace that conversation.
It should make the conversation more efficient.
2. AI-assisted documentation and clinical summaries
Documentation is essential to medical practice. It provides continuity of care, communicates clinical reasoning, supports handover, and creates an important medico-legal record. But documentation also consumes considerable physician time.
This is particularly evident when clinicians have to enter the same information repeatedly into different forms, systems, and notes.
AI-assisted documentation could reduce this burden by generating a structured draft from the clinical encounter, which the physician can review, correct, and approve. But the potential goes beyond generating a consultation note.
Imagine an ED system that continuously maintains a summary of what has happened to a patient since arrival.
The summary could include:
- Presenting complaint and relevant history
- Important examination findings
- Investigations and their results
- Imaging performed and reported findings
- Treatments already given
- Response to treatment
- Specialist consultations
- Current clinical status
- Pending investigations
- Pending tasks
- Plan of care
A clinician or nurse joining the patient’s care several hours after arrival could review the summary instead of reconstructing the entire sequence of events from multiple notes. During handover, this could become particularly useful.
The outgoing physician could review the automatically generated summary, verify it, and hand over the important clinical issues. The incoming team would have a reference they could return to at any point during the shift. This is the same ground that ambient AI scribes in the emergency department are starting to cover.
The goal is not to eliminate clinical documentation or clinical judgment. It is to reduce the amount of time clinicians spend manually reconstructing information that already exists.
3. Real-time hospital capacity and workflow management
Bed availability is another major source of administrative work in the ED. A physician may decide that a patient requires ICU admission, but that decision does not immediately result in an ICU bed.
Someone still needs to find out whether a bed is available.
The same applies to ward beds, HDU beds, isolation rooms, ventilators, and sometimes even transport facilities.
A real-time hospital capacity system could display this information centrally. If an ICU bed becomes available, the relevant team could know immediately.
If a patient is waiting for a ward bed, the system could track the request and current status. If a transfer is required, the workflow could potentially integrate receiving hospitals, bed availability, ambulance availability, required equipment, and transfer documentation. The physician would still decide where the patient needs to go and whether transfer is appropriate.
The technology would simply reduce the administrative effort required to make that happen.
The goal isn’t to replace the human part of medicine
There is an important distinction here.
Not every part of a physician’s workload should be automated.
Communication with patients and families is one example.
Explaining a diagnosis, discussing prognosis, taking informed consent, answering questions, and helping a family understand difficult decisions are fundamentally human interactions. Technology should support these conversations, not replace them.
Similarly, clinical decisions cannot simply be delegated to an algorithm. A patient with septic shock may require fluid resuscitation, but there is no single fluid volume that is appropriate for every patient – cardiac function, renal function, pulmonary status, and the response to initial fluids all influence the decision.
A patient presenting with acute ischemic stroke within the thrombolysis window may appear to be an obvious candidate for treatment, but the physician still has to consider the clinical presentation, imaging findings, blood pressure, anticoagulant use, bleeding risk, time of symptom onset, and other eligibility criteria rather than acting on a single automated flag. Likewise, a patient with respiratory failure may meet criteria for intubation, but the decision depends on the cause of respiratory failure, prognosis, airway patency, mental status, and response to initial treatment.
AI can bring the relevant information together; the clinical decision still belongs to the physician.
An AI system can summarize a patient’s previous records, but the physician still has to interpret what those records mean; it can identify an available ICU bed, while the physician decides whether the patient actually requires intensive care; and it can draft a clinical note, while the physician remains responsible for ensuring that the documentation is accurate. The objective is not to make technology the doctor. It is to make technology the infrastructure that allows the doctor to function better.
The physician shouldn’t be the system’s messenger
This is perhaps the simplest way to think about the opportunity.
A highly trained emergency physician should not have to spend several minutes finding out whether a bed exists, searching through paper files for an old ECG, checking repeatedly for a pending investigation, or relaying the same patient history to multiple departments.
These are not necessarily clinical problems. They are often information and workflow problems.
Technology can make the relevant information available to the right person at the right time, while leaving the interpretation, communication, and clinical decisions to the people responsible for them.
That distinction matters.
The goal is not to remove humans from the healthcare system.
It is to remove unnecessary friction from the way information moves through it.
Bringing the physician back to the patient bedside
The problem with administrative work is not that it lacks value. Documentation, bed allocation, interdepartmental coordination, access to previous medical records, and investigation tracking are all essential to delivering good patient care. The problem is that much of this work is being performed manually by people whose primary skill is clinical decision-making and immediate resuscitation.
And this brings us back to the “brain juice” we started with.
Every emergency physician has a finite amount of cognitive energy during a shift. That reserve should be spent where it has the greatest value – recognizing the deteriorating patient, solving the difficult diagnostic problem, deciding when to intubate, interpreting the ECG that doesn’t quite fit, performing the procedure, speaking to the family, or deciding what needs to happen next.
It shouldn’t be unnecessarily depleted by repeatedly searching for a CT report, calling multiple departments with the same patient history, checking whether a potassium result has arrived, or manually reconstructing what happened to a patient six hours ago. If technology can remove even a few minutes of administrative work from every patient, the cumulative effect across a busy emergency department could be substantial. The future ED should therefore not be one where AI replaces emergency physicians. It should be one where the physician spends less time searching, typing, calling, checking, and coordinating – and more time assessing, treating, communicating, and making decisions.
The goal isn’t to give the physician more work with better technology. It is to preserve more of their “brain juice” for the patients who need it.
Technology should not make the physician the system’s messenger.
It should make the system work for the physician.
Where Cagnea fits
Cagnea works on the part of this problem that sits closest to the physician, which is the documentation itself. It listens to the encounter and drafts the note, so the clinician reviews and signs rather than types. The clinical judgment, the conversation with the patient, and the decision about what happens next all stay where they belong.
You can see how it works, or what it costs.
