Which Indian languages does KiraDoc support?
Thirteen languages are in production today: Hindi, Tamil, Telugu, Bengali, Marathi, Gujarati, Kannada, Malayalam, Punjabi, Odia, Urdu, Assamese and Maithili. Language is detected automatically per speaker, so there is nothing to configure before a consultation — you just start recording and speak normally.
Does it handle Hinglish and code-mixed speech?
Yes — and this is the normal case, not an edge case. Indian OPD consultations are overwhelmingly code-mixed: a regional language carrying English medical terminology ("chest mein pain hai, ECG karwa lo"). KiraDoc transcribes each segment in its native script and keeps the English clinical terms intact, because that mixture is what doctors actually dictate.
Is the patient's audio recording stored?
No. The uploaded audio is deleted from disk as soon as the transcription pipeline finishes — it is removed in a guaranteed cleanup step that runs even if processing fails. What persists is the transcript and the clinical note, held in an India-hosted PostgreSQL database with column-level encryption on patient identifiers. Audio is a transient input, never an archive.
Does the doctor still review the note?
Always. KiraDoc drafts; the doctor approves. The note is a draft until a clinician signs off on it, and every clinical statement in it is traceable back to the line of transcript it came from, so you can verify rather than trust. Where the model is not confident about something clinically load-bearing — a drug name, a dosage — it flags the uncertainty instead of inventing a plausible answer.
How long does a consultation take to process?
Seconds for a typical OPD consultation. Submission is asynchronous: you get a job ID back immediately and the app shows live stage-by-stage progress (preprocessing → speaker separation → language detection → transcription → note generation), so nothing blocks your screen while it runs.
Where is patient data stored, and is it compliant with Indian law?
Clinical data stays in India. Consultation audio, transcripts, notes and patient identifiers are held in a database hosted in the ap-south-1 (Mumbai) region, with column-level encryption on identifiers. Anonymous product analytics — page views and feature usage, never patient content — is the one exception and is processed outside India by our analytics provider. The product is built around India's DPDP Act 2023 and the Telemedicine Practice Guidelines 2020 — including the doctor-side consent model those rules actually contemplate — rather than a retrofitted US HIPAA workflow. For hospitals with data-sovereignty requirements, KiraDoc can also be deployed fully on-premise so that no audio or note leaves your building.
Does KiraDoc work with ABDM / ABHA?
ABHA patient linking and FHIR R4 record assembly are built, with clinical terms mapped to standard vocabularies before packaging rather than raw text wrapped in a FHIR envelope. Formal ABDM milestone certification is in progress — we are not claiming certified status until it is granted. Once certified, notes generated in KiraDoc can flow to any ABDM-compliant hospital system without per-vendor integration work.
Can it integrate with the HIS or EMR we already use?
Yes, two ways. Everything KiraDoc does is exposed over a documented REST API (
api.kiradoc.in/docs), so an existing system can submit audio and pull back structured notes directly. For hospitals, we also support FHIR R4 output so notes land in standards-compliant systems without custom glue. Email us about your specific HIS and we will tell you honestly whether it is a week of work or a month.
What does it cost?
There is a free tier covering 50 consultations a month, intended for a doctor evaluating it properly rather than a time-limited trial. The Clinic plan is ₹2,999/month for up to 1,000 consultations and 5 practitioners. Hospitals and chains needing on-premise deployment or HIS integration are priced individually — see
pricing.