What changes with automatic transcription
An electronic record with AI records the visit (in person or by telehealth), transcribes the audio and organizes the content into the fields of a clinical history — chief complaint, history of present illness, current medications, physical exam. The doctor examines and talks normally; what changes is not having to type during the visit.
That solves a concrete problem: looking at a screen instead of the patient changes the quality of the consultation, and leaving it all for later turns into backlogged paperwork — or simply doesn't get done with the detail it should.
What the AI fills in, and what it never fills in
The AI structures what was said: symptoms reported, medications (always by scientific name), history mentioned in conversation. It doesn't deduce, doesn't calculate and doesn't decide on its own.
- Diagnostic hypothesis, ICD-10 and treatment plan are always left blank by default — fields reserved for the doctor
- The AI never suggests a diagnosis or a treatment, even when the model could plausibly "guess" from context
- Medications mentioned by brand name ("Tylenol") appear in the document by scientific name ("Acetaminophen")
- Every generated text needs to be reviewed by the doctor before use — that applies to any AI-generated information, not just in medicine
In person or telehealth, the process is the same
In the practice, recording happens through the computer's microphone. In telehealth, the system captures the microphone (your voice) and the call's sound (patient's voice), on any platform — WhatsApp Desktop, Google Meet, Zoom, Teams. It's also possible to attach audio already recorded on a phone, if the visit happened elsewhere.
Privacy: what stays on your computer and what doesn't
Each team member's voice is biometric data and stays stored locally only, never in the cloud. There's no patient registration — identification happens through the content of the conversation itself, and files are organized only by date.
To transcribe and structure the visit, the audio and text need to be processed by AI providers — that requires an internet connection. Recording itself works offline; only generating the final record depends on network access.
Want to see this working in your clinic?
Talk to Marcelo