Multilingual AI scribe: documenting consultations in any language

In short

A multilingual AI scribe listens to a consultation held in any supported language and drafts a structured clinical note, such as a SOAP note, in the language the clinician chooses. When it runs alongside real-time medical interpretation, the clinician and patient can each speak their own language while the note is drafted from the same conversation. The clinician then reviews, edits and approves the note before it is filed.

Why most AI scribes struggle with non-English consultations

Ambient AI scribes have been built mainly for consultations held in English between an English-speaking clinician and an English-speaking patient. Clinical reality is different. Many services see patients every day who are more comfortable in another language, and those encounters are often the hardest to document:

• Three-way conversations. With a phone or in-person interpreter, the scribe hears two languages and three voices, and has to work out what the patient actually said.
• Mixed-language speech. Patients and families often switch between languages mid-sentence, especially for symptoms, medications and family history.
• Notes written afterwards. When the scribe cannot follow the conversation, the clinician writes the note from memory at the end of the session, which is exactly the burden the scribe was meant to remove.
• Equity of care. Patients who do not speak English end up with shorter, less detailed records than everyone else.

How a multilingual ambient scribing workflow works

With Ozana AI Interpreter and Ozana AI Scribe running on the same engine, a consultation follows five steps:

1. Start the session and set the note language. The clinician chooses the output language for the note, for example English, before the consultation begins. It runs in the browser, so there is nothing for the patient to download.
2. Speak naturally. The clinician speaks in their language and the patient answers in theirs. The Interpreter translates voice-to-voice in both directions in under 200 milliseconds, so the conversation keeps its natural pace.
3. The note drafts itself. The Scribe listens to the same live stream and drafts the SOAP note while the consultation is still in progress, rather than working from a separate recording.
4. Review and approve. When the encounter ends, the draft is ready in seconds. Nothing is filed automatically: the clinician reviews, edits and approves every note.
5. Nothing is kept. Audio and transcripts are processed in memory and discarded. Only the clinician-verified note leaves the session, sent to the record system or downloaded by the clinician.

Three example consultations

A Spanish-speaking patient in primary care

A GP sees a patient with a two-week cough who is most comfortable in Spanish. The GP asks questions in English, the patient replies in Spanish, and both hear the other in their own language. At the end of the ten-minute appointment, the GP has an English SOAP note covering history, examination findings, assessment and plan, ready to review before the next patient.

An Arabic-speaking patient in urgent and emergency care

In an emergency setting there is rarely time to book an interpreter. The clinician can start a session on any device, take a history in Arabic and English, and leave with a structured note and handover summary that the rest of the team can read.

A Mandarin-speaking family in an outpatient clinic

A consultant reviews an older patient who speaks Mandarin, with a family member present. The note can be produced in English for the hospital record, or in the language of the consultation if that suits the clinician, so details about medications and next steps are captured once, accurately.

What a SOAP note looks like when languages differ

The structure does not change with the language of the consultation. The note follows the format clinical teams already use:

• Subjective: the patient's account, in their words, translated into the note language.
• Objective: examination findings and observations shared by the clinician.
• Assessment: the clinician's working diagnosis and reasoning.
• Plan: treatment, prescriptions, referrals and safety-netting advice.

Ozana AI Scribe also adds a problem list and next steps, so follow-up is clear for whoever picks up the record.

Choosing the best medical AI scribe for non-English consultations

If a meaningful share of your patients speak a language other than your own, ask these questions of any AI scribe before you adopt it:

• Language coverage: which languages does it document, and does that match your local population?
• Note language: can the note be produced in a different language from the consultation?
• Interpretation: does it help the clinician and patient understand each other, or does it only document?
• Clinical vocabulary: is it tuned for symptoms, medications, dosages, procedures, consent and safeguarding?
• Clinician control: does every note require review and approval before it is filed?
• Data handling: are recordings or transcripts stored, and if so, where and for how long?
• Governance: how does it align with HIPAA, GDPR and, in the UK, NHS DTAC?

Why combine interpretation and scribing

Interpretation and documentation are usually separate services: an interpreter for the conversation and, if one is used at all, a scribe for the note. Ozana AI brings them together on one secure engine. The same live stream that powers the interpretation feeds the note, so there is one consultation and one record, with no recordings or transcripts stored at rest. You can read more about how this works on the technology page and our compliance posture.

Clinical safety reminders

• The clinician remains responsible for the content of every note and should read it in full before approving.
• Follow your organisation's policy on when a professional interpreter must be used, for example for some consent and safeguarding conversations.
• Check translated medication names, doses and allergies with the patient during the consultation.

Next steps

To see a multilingual consultation documented end to end, book a demo with the team, or read about how Ozana AI is used in hospitals, GP practices and ambulance services on the use cases page.