Why I Built NEA Scribe: A Practitioner's Case for Presence in Healthcare AI

I.

The Problem No One Could See

I have sat in rooms where people were dying. I have stood at the threshold of ICU bays, listening to families try to say in five minutes what they never found time to say over five decades. I have watched nurses and physicians move through these moments with practiced efficiency — and then I have watched chaplains do something none of the other clinicians were trained to do: stay. Not to fix anything. Just to stay.

That staying is what clinical chaplaincy is, at its most essential. And for years — through my own training, through unit rotations, through the slow and difficult accumulation of what CPE supervisors call "competence" — I understood that the work happened in the room, in the silence, in what passed between two human beings when one of them was confronting the fact that they were going to die.

What I did not understand, for a long time, was how catastrophically the documentation failed to capture any of it.

The clinical note written at 6pm for an encounter that happened at 11am would read: "Patient visited. Expressed anxiety related to prognosis. Spiritual support offered. Patient verbalized appreciation." Fourteen words. The encounter itself had lasted forty minutes. It had involved tears, silence, a story about a sister who died of the same cancer twelve years earlier, a question about whether God punishes people, and a moment — brief, unrepeatable — where the patient said something that I will carry with me for the rest of my life.

None of that appeared in the note.

I told myself this was acceptable. Every clinician loses something in documentation; that's the nature of clinical work. The note captures what matters for the medical record. The rest lives in your memory, your supervision sessions, your own processing. But over time — over hundreds of encounters, across multiple hospital systems — I began to understand that this framing was wrong. It wasn't that documentation captured the clinically relevant portion and discarded the rest. It was that the documentation tools chaplains had access to were so inadequate for their actual work that chaplains had unconsciously learned to document less than they noticed, less than they engaged with, less than they actually knew about the patient's spiritual and existential situation.

This is the subtler form of the problem, and it took me years to name it clearly. When your tools can only capture a narrow slice of what you observe, you start unconsciously curating your observations to fit the tool. You arrive at the bedside and you notice the forty things you always notice — the family dynamics, the grief topology, the theological frame the patient is using to make sense of their illness — but you have already, somewhere in the back of your mind, discarded thirty-five of them because you know they won't survive the documentation process.

The result is a clinical record that systematically understates the spiritual complexity of hospital patients. And a profession that, consequently, struggles to demonstrate its impact, justify its staffing, or pass meaningful knowledge between care team members.

I am not the first person to notice this. Chaplains have been saying it for decades. What changed, for me, was when I started to believe that the technology finally existed to do something about it.

II.

What Presence Actually Looks Like

Before I explain what NEA Scribe does, I need to explain what spiritual care actually is — because there is enormous confusion about this, even within healthcare systems, and that confusion is part of what makes building tools for chaplains so difficult.

Spiritual care is not religious care, though it includes religious care when religion is part of a patient's life. Spiritual care is the practice of attending to a person's deepest questions about meaning, identity, suffering, and connection — questions that become acute when illness strips away the routines and certainties that normally buffer us from those questions. A patient who has never been religious in their life may, facing a serious diagnosis, find themselves asking: Why did this happen to me? What was my life for? Will anyone remember me? Is there anything after this? These are spiritual questions. And the clinician trained to engage them — without judgment, without an agenda, without needing to resolve them — is the chaplain.

The framework at the center of NEA Scribe reflects how that engagement actually unfolds. We call it the NEA framework: Noticing, Engaging, Accompanying. These are not steps in a protocol. They are the three dimensions of any chaplaincy encounter, the three registers in which the chaplain is simultaneously operating.

What I want to illustrate is why this framework — and the encounters it describes — resists standard clinical documentation.

Consider three encounter archetypes that every hospital chaplain knows well. The first: the patient who will not speak to anyone else. I have met patients — sometimes in oncology, sometimes post-surgical, sometimes in long-term care — who have stopped communicating with their nurses, their physicians, even their family. The care team is worried, but the patient won't explain why. In my experience, these patients often will speak to the chaplain, and what they say often reveals that they have not stopped communicating because they are unable to — they have stopped because they feel that no one in the clinical environment has time for what they actually need to say. The chaplain's role, in that encounter, is not to relay information to the team (though sometimes, with consent, pieces of what emerges do need to reach the team). It is to be the presence that makes speech feel safe again.

How do you document that? What goes in the note? "Patient previously non-communicative; engaged in conversation with chaplain" captures the observable fact but erases the meaning. What was said? What shifted? What is the spiritual portrait of this patient that the care team needs to understand in order to continue providing good care? These are the questions that chaplaincy documentation is supposed to answer — and almost never does, because the tools provide no structure for them.

The second archetype: the family holding vigil. The patient is dying — actively, in the clinical sense — and three generations of family are arranged around the bed in various states of disintegration. The chaplain's presence in that room is not primarily for the patient, who may no longer be conscious. It is for the people who will carry this death with them for the rest of their lives. What the chaplain notices — who is closest to the bed, who is standing apart, which relationships seem strained, what each person seems to need in this moment — is clinically significant information. Not for the ICU physician, who is managing the physiological process of dying, but for the care team as a whole: the social worker who will do bereavement follow-up, the nurse who will continue to interact with the family through the final hours, the chaplain's own future self who may encounter members of this family again in other contexts.

The third archetype is the one I find most instructive, because it is the one that most directly exposes the limitation of generic documentation tools. The secular skeptic. The patient who opens the conversation by saying, "I'm not religious, so I'm not sure why they sent you," and then — over the next hour — asks every question about meaning, mortality, and legacy that any religious person has ever asked. These conversations are among the most technically demanding in chaplaincy, because the chaplain must meet the patient in their own frame while still offering the depth of presence that spiritual care requires. And they are among the most underdocumented, because the EHR note says "patient identifies as non-religious" and stops there, as though spiritual care requires a religious entry point.

These encounters are not retrospectively reconstructible from memory. The specific texture of what happened — who said what, what silences meant what, what the chaplain noticed and chose to respond to and why — begins to fade within hours. By the time a chaplain at the end of an eight-hour shift sits down to document the four or five encounters they have had, the living detail is already gone. What survives is an outline. And outlines do not convey clinical significance.

III.

Why Existing Tools Fail

I spent a significant period of time, before building anything, trying to understand why the documentation problem had persisted so long. Chaplains are not unaware of it. CPE programs are not unaware of it. Healthcare systems that employ chaplains are increasingly measuring their spiritual care departments and wondering why the data is so thin. So why had no one fixed it?

The first and most significant reason is that electronic health record systems were designed for physicians. The clinical documentation architecture of nearly every EHR in use — the problem-oriented medical record, the SOAP note structure, the billing-code taxonomy — was built to capture what physicians do. When chaplains were integrated into EHR systems, they were given a small slice of that physician-centric architecture and told to make it work. The result is like asking a composer to describe their symphony in the format of a financial ledger. You can fit some information in. Most of the important information doesn't have a field.

The second reason is voice dictation tools. These were the obvious solution — clinicians in other specialties have used them for years, and they solve the speed problem. But voice dictation is not the same as voice comprehension. A dictation tool records what you say. It does not understand chaplaincy vocabulary, recognize the significance of a theological statement the patient made, or organize the content of a complex spiritual assessment into the structure that clinical documentation requires. Chaplains who have tried generic dictation tools report that they feel like talking into a void: the transcript is accurate, but it still requires the same amount of post-processing to turn into a useful note.

The third reason is the most systemic, and the one that has received the least attention. Clinical chaplaincy has an administrative overhead problem that compounds every other problem. In many hospital systems, chaplains are responsible not only for direct patient care but for on-call coverage coordination, care plan participation, interdisciplinary team communication, bereavement outreach, and their own continuing education. The documentation workload does not exist in a vacuum — it competes with everything else a chaplain needs to do. When documentation is difficult, slow, and cognitively demanding, chaplains rationally make the same calculation that every clinician makes when their documentation tools are inadequate: they do the minimum required and redirect the saved time toward the work that feels most valuable, which is direct patient contact.

This is not negligence. It is a rational response to a broken system. But it creates a vicious cycle: thin documentation makes chaplaincy look less clinically significant than it is; departments that look less significant get fewer resources; fewer resources mean chaplains are more overstretched and document even less; and so on. The documentation problem is not a peripheral issue for the profession. It is near the center of why chaplaincy continues to be underfunded, understaffed, and underestimated in clinical settings.

IV.

The Design Principle That Changed Everything

When I began working on what would become NEA Scribe, the first question I had to answer was: what would it mean to build a documentation tool that was actually designed for chaplains, rather than adapted from something designed for someone else?

The answer I kept returning to was deceptively simple: it should feel like talking to a supervisor.

Every chaplain trained in Clinical Pastoral Education knows what supervision feels like. You describe what happened in an encounter — in your own words, with the texture and detail intact — and your supervisor helps you understand what you observed, what it meant, what it tells you about the patient's spiritual situation, and what it tells you about your own practice. The supervision process transforms raw experience into clinical knowledge. And it does this through conversation, not through form-filling.

That insight drove every design decision we made. Voice-first, ambient capture means that a chaplain, immediately after leaving a patient's room, can spend two minutes speaking naturally about what just happened. Not dictating a note — talking. Describing. Remembering aloud. The system listens, understands, and then produces a structured clinical note that reflects both what was said and what it means in the context of spiritual care practice.

This required building something we call chaplaincy vocabulary comprehension — the capacity to understand the specific conceptual vocabulary that trained chaplains use. Terms like "spiritual distress," "meaning-making," "lament," "theodicy," "anticipatory grief," "existential crisis," and dozens of others have precise meanings in the context of spiritual care that generic language models do not recognize or handle well. When a chaplain says "the patient is in a place of deep lament about their diagnosis," NEA Scribe understands that this is a clinically significant statement about the patient's spiritual state — not a generic expression of sadness — and treats it accordingly in the generated note.

The NEA score — the numerical summary that appears at the top of each encounter note — was one of the features I was most uncertain about and have since become most convinced matters. It quantifies the depth of engagement across the three dimensions of the framework: how much the chaplain noticed, how deeply they engaged, how sustained the accompaniment was. The score is not a performance metric. It is a reflective tool. When a chaplain sees their encounters over time and notices that their Noticing scores are consistently high but their Accompanying scores are lower, that is information. It points toward a pattern in their practice that they can bring to supervision, work on in CPE, and consciously address.

This brings me to private mentor mode, which is the feature I am most proud of and which has no equivalent in any other clinical AI tool I am aware of. Every chaplain needs reflective supervision. Not every chaplain has consistent access to it — particularly those in smaller hospitals, rural settings, or systems where CPE supervision is infrequent. Private mentor mode is an AI conversation partner, built on the same chaplaincy-specific knowledge base, that offers exactly what a good supervisor offers: a space to process what happened, explore what you noticed, examine what you felt, and think about what you would do differently. The conversation is private. It never enters the clinical record. It is not peer review and it is not therapy. It is the reflective container that every chaplain deserves and most chaplains do not reliably have.

V.

What Presence-Centered AI Means

There is a question that I hear, almost every time I describe this tool to chaplains who have not yet tried it: does AI belong anywhere near this work?

I take the question seriously. It is not naivety about technology; it is a professional and ethical instinct that the most human elements of clinical care need to be protected from the particular kind of abstraction and reduction that technology tends to produce. When people hear "AI in chaplaincy," they often imagine a chatbot offering scripted reassurances to patients who should instead be talking to a human being. That would be a catastrophic application of the technology, and it is not what we are building.

The distinction I keep returning to is this: AI should make chaplains more present, not less. The enemy of presence is not technology — it is cognitive load. When a chaplain is in a patient's room with the documentation task already forming in the back of their mind, they are not fully present. When they are mentally composing the note while the patient is speaking, they are not fully listening. When they leave the encounter early because they have four more visits and two notes still outstanding, the patient loses something real.

NEA Scribe is designed to eliminate the documentation tax on presence. If the two minutes after leaving a room are sufficient to produce a complete, well-structured clinical note, then the chaplain can give the room all of their attention while they are in it. The encounter does not have to compete with the documentation. The presence does not have to be rationed.

There is a broader question implicit here about what AI will do to healthcare. The dominant anxiety — justified in many contexts — is that AI will be used to substitute for human contact: to replace conversations with chatbots, to route patients away from clinicians and toward algorithms. This anxiety is well-founded. Poorly designed healthcare AI can, in fact, reduce human presence in medicine.

But I am increasingly convinced that there is an equally important category of healthcare AI that does the opposite: that takes on the work that was always extracting time from clinical relationships, and hands that time back to clinicians. Documentation is the most obvious example. The average physician spends more time documenting than they spend in direct patient contact. The administrative burden of clinical work is not incidental to the care crisis — it is a central driver of it. Clinicians who are burned out, overstretched, and documentation-exhausted are less present to their patients, less available for the kinds of conversations that actually heal people, and more likely to leave the profession.

For chaplains, this dynamic is especially acute, because presence is not merely part of what they do — it is the entire mechanism of care. You cannot offer spiritual accompaniment while you are mentally composing a clinical note. The quality of the relationship is the intervention. If documentation overhead degrades the quality of presence, documentation overhead degrades the clinical outcome.

AI that understands this — that is designed explicitly to protect presence rather than to replace it — is what NEA Scribe is trying to be. I think this design philosophy is the right one not just for chaplaincy but for every clinical specialty, and I think the profession that has always made presence its central practice is exactly the right place to demonstrate what presence-centered AI can look like.

VI.

An Invitation

I want to be specific about who this is for, because "AI tool for healthcare" covers an enormous range of people with very different needs.

If you are a hospital chaplain — board-certified or in the process of becoming so — this is designed for your daily practice. It handles the documentation so you can handle the care. If you are seeing eight patients a day and spending two hours on notes, we want to give you those two hours back. If you are struggling to demonstrate the clinical impact of your department to a skeptical administrator, we want to give you the data and the language to make that case.

If you are a CPE supervisor, the mentor mode and NEA scoring open new possibilities for how you structure supervision. Patterns in a student's encounter scores reveal things that verbal reporting alone often doesn't. When a supervisee says "I felt present in that encounter" but their Noticing score is consistently low, you have something concrete to work with. That conversation — grounded in specific data from specific encounters — is, in my experience, more productive than the same conversation grounded only in recollection.

If you are a healthcare system leader evaluating spiritual care programming — a CNO, a chaplaincy department director, a VP of Patient Experience — the documentation infrastructure that NEA Scribe creates begins to answer questions you have probably been unable to answer: What is the depth and consistency of spiritual care across your institution? How does the intensity of chaplaincy engagement correlate with patient satisfaction, length of stay, or family outcomes? Where are the gaps in coverage? These are questions that matter for staffing decisions, accreditation standards, and institutional reputation — and they have historically been unanswerable because the data simply didn't exist in structured form.

If you are a TMC member, a healthcare ethics consultant, or a researcher in spiritual care, the dataset that NEA Scribe creates — in aggregate and de-identified form — represents something that has never existed before: a structured, framework-consistent record of spiritual care encounters at scale. The research implications alone are significant enough that I expect a meaningful portion of our future development to be shaped by what researchers discover in that data.

Free to start means exactly that: you can create an account, connect your voice, and complete your first encounter note today, without a credit card, without a demo call, without a procurement process. I built the free tier to be genuinely useful — not a hobbled preview — because I believe the tool needs to prove itself in practice before it asks for commitment. The upgrade path is there when you need the full clinical suite, the department analytics, the EHR integration, and the priority support. But the place to start is to try it on one real encounter and see whether the note it produces reflects what actually happened in the room.

I started this essay by describing an encounter that produced a fourteen-word clinical note. I have thought about that particular encounter many times over the years — about the patient, about what was said, about what I missed and what I managed to offer. The documentation failure was not, in isolation, catastrophic. The patient received good care. The encounter mattered, whether or not the note reflected it.

But multiply that encounter by the tens of thousands of chaplaincy visits that happen every day across every hospital system in the country. Multiply it by the careers of chaplains who never had access to tools adequate to their work. Multiply it by the patients whose spiritual complexity was never captured in the clinical record, never informed the care plan, never shaped how the next clinician approached the room.

The aggregate loss is enormous. Not catastrophic in any individual case, but enormous in what it has cost the profession, the patients, and the evidence base for spiritual care as a clinical discipline.

That is what NEA Scribe is trying to recover. Not the specific encounters — those are gone. But the practice of capturing them fully, honoring them as clinical data, and using them to build the case that spiritual care is not a soft complement to medicine but a core dimension of what healing is.

The work matters too much to let documentation get in the way. Come try it.