Video visits in MindFlow use Zoom as the underlying video platform. Sessions are created, tracked, and linked to appointment records automatically. The patient receives their join link without needing a Zoom account.
The integration uses Zoom's Server-to-Server OAuth, clinic credentials are stored securely in the platform settings and used to create and manage meetings on behalf of the clinic. No individual Zoom user accounts are needed for staff.
When a video appointment is confirmed, a Zoom meeting is created automatically and the join link is stored against the appointment. Zoom sends real-time event notifications (participant joined, left, recording completed) via webhook, and the appointment record is updated accordingly.
Patients join via a browser using the Zoom Web SDK, no Zoom app installation needed. The clinic admin can configure whether to offer the browser-based or native-app join link depending on their patient population.
All inbound Zoom webhook events are verified against Zoom's HMAC signature before any processing occurs. Events with invalid signatures are rejected and logged.
The recording interface will not start until the patient's consent has been logged in the system. This is enforced at the server level, a recording upload without a consent record is rejected, not just hidden in the UI.
Audio can be captured directly in the browser (using the Web MediaRecorder API, chunked and uploaded as the session progresses) or via Zoom's cloud recording feature. Both paths feed into the same transcription pipeline.
Completed audio is sent to Sarvam AI for transcription. The transcript is stored against the session and made available to the clinician for note drafting. Indian languages are supported alongside English.
Recording files are stored encrypted. Access requires authentication and is logged. Files are never accessible via a direct or guessable URL.
The demo comes pre-loaded with real sample data, so you can click through actual appointments, patients and prescriptions.