Generate a SOAP note in 30 seconds — with no signal on the ward

It is 7:14 on a Tuesday morning. You are standing at the foot of a patient's bed on a ward where the hospital network blocks personal mobile data and the Wi-Fi on your phone shows one bar that comes and goes. You finished clerking this patient twenty minutes ago. The ward round starts in six minutes. The SOAP note you need to write is currently living in your short-term memory, competing with the three patients you saw before this one.

This is the documentation reality of clinical training in 2026. It has not changed meaningfully in thirty years, despite every other aspect of medicine advancing. The tools available to address it have.

What a SOAP note is and why it matters

A SOAP note — Subjective, Objective, Assessment, Plan — is the standard clinical documentation format used across medicine, nursing, and allied health. It structures a clinical encounter into four sections:

• Subjective: What the patient reports — symptoms, history, complaints in their own words
• Objective: Clinical findings — observations, vital signs, examination findings, test results
• Assessment: Clinical interpretation — diagnosis or differential diagnosis
• Plan: Next steps — investigations, treatments, referrals, follow-up

The format exists because clinical communication needs to be fast to write, fast to read, and unambiguous. A well-written SOAP note in a patient record allows the next clinician who sees this patient — at 2am, two weeks later, in a different hospital — to understand what happened and what was decided.

The problem is that writing a good SOAP note requires time, concentration, and accurate recall. All three are in short supply at 7:14 on a Tuesday morning on a ward where the next patient is already waiting.

The documentation burden in clinical training

Research consistently shows that junior doctors spend between 30–40% of their working time on documentation. For a 12-hour ward shift, that is 3.6–4.8 hours spent typing, dictating, or writing — time not spent with patients, not spent learning from senior colleagues, not spent thinking clinically.

The British Medical Association has documented the administrative burden on doctors as one of the leading contributors to burnout. Documentation that happens retrospectively, from memory, at the end of a long shift, is both inefficient and vulnerable to error. Details blur. The specific words a consultant used to frame an assessment at 8am become fuzzy by the time you're typing the note at 7pm.

AI documentation tools have begun to address this. Enterprise clinical scribes like Heidi Health (currently $150/user/month) and Nabla (approximately $119/month) process consultation audio and generate structured clinical notes. They work well in private practice settings — GP surgeries with a dedicated office space, stable broadband, and a billing rate that justifies the subscription cost.

They are largely inaccessible to NHS junior doctors, medical students, and allied health professionals who need the same capability but don't have $150/month to spend on a personal work tool, and whose ward environment has no reliable internet for the tool to use.

Why existing AI tools don't work on the ward

There are three structural reasons why cloud-based clinical AI tools fail in the NHS ward environment:

Connectivity. Hospital wards regularly block personal mobile hotspots. Trust network access is controlled and often unavailable to personal devices. The NHS Digital baseline for ward connectivity is improving, but in practice, a junior doctor's personal iPhone on a ward in 2026 is often operating with no usable data connection.

Data governance. Patient audio is Article 9 special category health data under UK GDPR. Transmitting it to a US cloud company's servers requires a Data Processing Agreement, explicit lawful basis, and patient consent that is almost never obtained in practice. A junior doctor who records a ward round on Otter.ai or any consumer cloud tool and uploads that audio to a US server has, in all likelihood, committed a data protection breach — even if the audio is deleted immediately after processing.

Procurement timescales. NHS IT procurement for new clinical software typically takes 18–24 months minimum. A junior doctor who starts their FY1 post in August cannot wait until 2028 for a trust-wide tool to be approved. They need something that works on their phone, today, within their own governance.

On-device processing resolves all three constraints simultaneously. No connectivity required. No data leaves the device — no DPA needed. No IT procurement process — it's an app on a personal device.

Generating a SOAP note in 30 seconds

Here is what the Kuulo workflow looks like during a patient clerking:

1. Before the encounter, open Kuulo and select the SOAP Note template
2. Tap record. Place the phone face-down on the desk or in your pocket
3. Clerk the patient: take the history, ask your questions, note their responses
4. The transcription runs in the background, entirely on your device
5. After the encounter, tap to generate the summary
6. Kuulo produces a structured SOAP note: patient-reported symptoms in Subjective, your clinical findings in Objective, the working diagnosis or differentials in Assessment, and the management plan in Plan
7. Review it. It takes 30–60 seconds. Correct any errors.
8. Copy it into your clinical system

The audio never leaves your phone. The patient's name, diagnosis, and history remain on your device. The GDPR position is clean.

What the SOAP note output actually looks like

The AI doesn't produce generic clinical language. The summary is built from what was said in the specific encounter. If the patient described their chest pain as "tight, like someone's squeezing" — that's what appears in the Subjective section, in quotes. If your examination found reduced air entry at the left base — that appears in Objective. The Assessment reflects the clinical reasoning you articulated. The Plan captures what you said you were going to do.

A 10-minute patient history that would take 15 minutes to type from memory takes 30 seconds to review and correct from an AI-generated draft. For a junior doctor seeing 15–20 patients on a morning ward round, this represents a fundamental change in how the shift flows.

Speaker diarization on the ward

Clinical conversations are multi-person. On a ward round, the consultant teaches, the registrar assesses, the FY1 clerks, the patient responds. In an MDT meeting, multiple clinicians contribute different perspectives.

Kuulo's on-device speaker diarization attributes statements to different speakers automatically. This means that in the transcript and summary:

• The consultant's teaching points appear attributed separately from the junior doctor's clinical observations
• The patient's reported symptoms are clearly distinct from the doctor's examination findings
• The registrar's assessment can be identified separately from the management plan discussion

For clinical portfolio evidence, this attribution is directly useful. A medical student who records a teaching ward round and produces an attributed summary has a significantly richer piece of portfolio evidence than notes taken manually.

The governance conversation made easy

When a clinical lead or Caldicott Guardian asks about AI documentation tools — and they will ask — the conversation needs to be brief and technically accurate.

For cloud tools: "The audio is sent to [company]'s servers for transcription, then deleted. They have a DPA. We'd need to review their DPIA and ensure lawful basis."

For Kuulo: "Everything runs on the clinician's own phone. Nothing leaves the device. There is no third-party data processor. No DPA is required. The audio sits on the phone until the clinician deletes it, in the same way a voice memo would."

The second conversation is shorter, cleaner, and requires no external audit to verify. The tool's architecture is its compliance evidence.

What this means for medical students on placement

Medical students on clinical placement face a documentation challenge that is both urgent and acute: they need to produce SOAP notes for their portfolio, demonstrate clinical reasoning in written form, and do so in an environment where they have the least time and the worst connectivity.

A SOAP note generated from a real patient encounter — reviewed, corrected, and annotated with the student's own clinical thinking — is a richer portfolio entry than a note reconstructed from memory after the placement day has ended. It captures what the patient actually said, what the student actually observed, and what the supervising clinician actually advised.

The template system also serves as a teaching scaffold. For a Year 3 student who has not yet internalized the SOAP structure, having the framework automatically applied to their clinical recordings makes the format visible and habitual. By Year 5, the structure is automatic.

The cost argument

Heidi Health charges $150/user/month — $1,800/year. Nabla charges approximately $119/month. These are compelling products for private practitioners with a billing rate that makes $150/month a rounding error.

For an FY1 doctor on a Band 1A foundation salary of approximately £32,000, a $150/month tool is not a rounding error. It is a monthly grocery bill.

Kuulo's core functionality is free to start. For a medical student or junior doctor who needs SOAP notes generated from ward encounters — offline, privately, with no procurement process — the economic argument is straightforward.

Using Kuulo appropriately in clinical settings

A few important practical points for clinical users:

Consent. The GMC's guidance on patient confidentiality requires that patients understand how information about them is being used. Before recording a consultation, explain that you're using a note-taking tool on your device. Most patients respond positively when told that nothing leaves the phone. If a patient declines, respect that immediately.

Purpose. Kuulo generates a draft clinical note. It is not a substitute for clinical judgment. Review the output, correct errors, and ensure it accurately reflects the encounter before copying it into any clinical system.

Trust policy. Some trusts have policies on personal device use in clinical areas. Check your trust's policy before using any personal device for clinical documentation.

These are not reasons to avoid the tool. They are the same common-sense considerations that apply to any documentation method. A doctor who uses Kuulo appropriately, with patient consent and within their trust's policies, is doing something categorically safer from a data governance perspective than a doctor who records a ward round on any cloud-based tool.

The ward has always been a difficult documentation environment. It now has a tool built specifically for it.