AI Assistant
The AI Assistant is a documentation scribe built into every consultation. It understands your practice type and patient context, and can populate consultation forms from natural-language instructions or a full session transcript.
The assistant is available as a floating chat panel during every consultation. Click the AI Assistant button (the robot icon) in the bottom-right corner of the consultation screen to open it.
!!! note "Scribe only" The AI Assistant records what you tell it. It does not suggest diagnoses, recommend medications, or provide clinical decision support of any kind. Always apply your own professional judgement to the information recorded.
Practice-aware language
The assistant automatically adapts its language and framing to your configured practice type. You do not need to use clinical codes or formal terminology — speak naturally and the assistant will use the right language for your discipline.
| Practice type | What it calls the person | What it calls the session | Notes style |
|---|---|---|---|
| Medical / Dental | Patient | Consultation / Appointment | SOAP |
| Physiotherapy / Chiropractic / Osteopathy / Podiatry | Patient | Appointment | SOAP (MSK focus) |
| Counselling / Psychology | Client | Session | Therapy note |
| Speech & Language | Client | Session | SLT session note |
| Massage / Reflexology / Complementary | Client | Session | Wellness session note |
Using text commands
Type a command in the input box and press Enter or click Send. The assistant acts on it immediately and confirms what was saved in one sentence.
You can be as brief or as detailed as you like. The examples below are grouped by what you want to record.
Vitals
BP 135/85, pulse 72
Temperature 37.4, weight 84kg, respiratory rate 16
Blood glucose 6.1 mmol/L, pulse 68 bpm
Record vitals: BP 120/80, pulse 74, temp 36.8°C, weight 71kg
Presenting complaints / Presenting concerns
Add a complaint: lower back pain radiating to the left leg, started 3 weeks ago
Presenting complaint is a productive cough for 5 days with fever
Client presenting with low mood and difficulty sleeping for the past month
Presenting concern: difficulty swallowing liquids, onset 2 weeks ago
Add complaint: right knee pain on weight-bearing, worse on stairs
Chief complaint: toothache upper right quadrant for 4 days
History
Past medical history: hypertension, type 2 diabetes, appendectomy 2015
Social history: non-smoker, drinks alcohol socially, works as a teacher
Family history: father had MI at 58, mother has breast cancer
Medication history: on metformin 500mg twice daily and ramipril 5mg
Client has a history of generalised anxiety disorder, previously worked with a therapist 2019–2020
Surgical history: knee replacement left knee 2021, no complications
Add history: patient reports seasonal allergies and previous episode of lower back pain 2 years ago that resolved with physio
Examination findings
Examination: abdomen soft, non-tender, no organomegaly, bowel sounds present
Chest clear to auscultation bilaterally, no wheeze or crackles
ROM: lumbar flexion 50%, extension 30%, lateral flexion restricted bilaterally
Straight leg raise positive at 40 degrees on the right
Oral examination: caries present on UL6, mild gingivitis upper arch
Client presented with flat affect, poor eye contact, and slowed speech
Add exam finding: tenderness on palpation of the right supraspinatus, pain arc 60–120 degrees
Cranial nerve examination normal, reflexes intact and equal bilaterally
Diagnoses and formulations
Diagnosis: lumbar disc herniation with radiculopathy
Add diagnosis: acute otitis media
Formulation: presenting with moderate depressive episode in the context of relationship breakdown and work stress
Diagnose: right shoulder rotator cuff tendinopathy
Diagnosis: type 2 diabetes mellitus, well controlled
Working diagnosis: functional neurological disorder
Clinical coding
The assistant can look up and apply codes from your configured coding system (ICD-10 or Read Codes) for items already recorded, or at the same time as recording them.
Code an item as you record it:
Add complaint: chest tightness — find the ICD-10 code
Diagnosis: essential hypertension — please code it
Code items already in the record:
Find and apply ICD-10 codes for all the diagnoses
Code the presenting complaint
Search for the ICD-10 code for type 2 diabetes and add it to the diagnosis
Look up a code without recording anything:
What is the ICD-10 code for Dupuytren's contracture?
Search read codes for plantar fasciitis
Prescriptions
Prescribe amoxicillin 500mg three times daily for 7 days
Add prescription: ibuprofen 400mg with food, three times daily, maximum 5 days
Metformin 1g twice daily with meals
Add: salbutamol inhaler 100mcg, 2 puffs as needed for breathlessness
Prescribe omeprazole 20mg once daily for 4 weeks
Add paracetamol 1g four times daily as needed, maximum 4g in 24 hours
Referrals
Refer to physiotherapy for lumbar disc rehab, routine
Urgent referral to cardiology — query arrhythmia
Refer to the community mental health team, routine, reason: moderate depression not responding to primary care treatment
Add a referral to podiatry, routine, for diabetic foot assessment
Refer to gastroenterology, urgent, suspected IBD
Refer to speech and language therapy, routine, for dysphagia assessment
Procedures
Procedure: wound dressing to left lower leg, clean and dry, re-dressed with non-adherent dressing
Performed cervical spine mobilisation grades III–IV
Joint injection: right knee intra-articular corticosteroid, 40mg triamcinolone
Procedure: ear syringe bilateral, wax successfully removed
Add procedure: Swedish massage, full body, 60 minutes
Tests and investigations
Order FBC, U&E, LFTs, HbA1c, fasting lipids
Request chest X-ray — query lower lobe consolidation
Order ECG
Add test: urine culture and sensitivity
Request MRI right knee with and without contrast
Order DEXA scan for osteoporosis screening
Clinical notes
Update notes: patient doing well, review in 3 months
Add to notes: patient was counselled on lifestyle modification including diet and exercise. No side effects from current medications. Plan to recheck HbA1c in 3 months.
Notes: client explored themes of low self-worth and fear of abandonment. Introduced CBT thought record. Homework: complete one thought record before next session.
Set notes to: SOAP note — S: patient reports 80% improvement in lower back pain. O: lumbar flexion 80%, SLR negative. A: resolving lumbar disc injury. P: continue home exercise programme, discharge after next session.
Drafting letters and documents
The assistant can draft referral letters and patient-facing documents from the information already in the consultation. Drafts are presented for your review — nothing is sent automatically.
Referral letters:
Draft a referral letter to cardiology
Write a formal referral letter to the gastroenterology team
Draft a concise referral letter to physiotherapy
The letter is pre-populated with the patient's details, current diagnoses, medications, history, and your practice information. Review and amend before printing or sending.
Patient medication instructions:
Write patient instructions for the amoxicillin prescription
Draft patient-friendly instructions for metformin 500mg twice daily with meals
Instructions for salbutamol inhaler use
Reviewing patient history
Summarise this patient's previous consultations
What were the main issues in this patient's last few visits?
Give me a summary of the client's recent sessions
What medications has this patient been on in previous consultations?
The assistant retrieves the last 5 consultations and produces a structured summary.
Session transcription mode
Paste the full text of a session recording and the assistant will process it in a single pass, automatically populating all active forms — complaints, history, examination, diagnoses, prescriptions, procedures, tests, referrals, and notes.
How to use it:
Prefix your message with Consultation transcript: followed by the transcript text:
Consultation transcript:
Doctor: Good morning, how have you been since last time?
Patient: Much better thanks, the back pain has settled a lot. Still a bit of stiffness in the mornings.
Doctor: Good to hear. Let me have a look at your range of motion...
[full transcript continues]
The assistant will:
- Extract vitals, history, complaints, examination findings, diagnoses, prescriptions, procedures, tests, and referrals from everything said during the session
- Populate only the forms that are enabled for your practice
- Write structured clinical notes appropriate for your practice type
- Confirm what was recorded and flag anything that may need your review
!!! tip "Audio recording" Use the built-in mic button in the AI Assistant panel to record the session in real time. When you click Stop, the transcript is automatically sent to the assistant for processing — no copy/paste needed.
Things the assistant will not do
The assistant is a documentation tool only. If you ask for something outside its role, it will say so and explain why.
| Request | Response |
|---|---|
| "What do you think the diagnosis is?" | Declines — explains it cannot provide clinical opinion |
| "Should I prescribe X or Y?" | Declines — this is clinical decision support |
| "Is this safe to prescribe?" | Declines — refer to your prescribing reference |
| "Write a referral and send it" | Drafts for review — never auto-sends |
| "Use the diagnosis form" (if disabled) | Explains the form is not enabled and suggests enabling it in Settings |
Tips
- You do not need to use exact field names. Say "add a note", "record a complaint", or "jot down the history" — the assistant understands natural language.
- Allergies are always in context. The assistant is aware of any allergies recorded on the patient's record and will include them automatically in referral letters and summaries.
- You can mix instructions in one message. For example: "BP 140/90, pulse 78. Add complaint: headache for 3 days. History: no previous migraines." The assistant will process all parts.
- Transcript mode works best with speaker labels. If your transcription software labels speakers (e.g. "Doctor:" / "Patient:"), the assistant can more accurately separate clinical findings from patient-reported symptoms.