EkaCare Voice Scribe: NHA picks Bengaluru firm's AI to write doctors' notes

The National Health Authority (NHA), the body that runs Ayushman Bharat, has chosen Bengaluru-based EkaCare as the technology partner for its ambient AI Voice Scribe, a tool that listens to a consultation between a doctor and a patient and turns it into a written health record. The selection was made public on Wednesday, 11 November 2026. EkaCare, operated by Orbi Health Private Limited, was picked through the NHA's Expression of Interest process for open-source voice-to-text tools that create digital health records under the Ayushman Bharat Digital Mission (ABDM).

Writing notes takes a large share of a doctor's time, and in busy public hospitals that work often means short or missing records. An "ambient" scribe is software that listens in the background while the doctor and patient talk, then drafts the note on its own. India adds a hard problem that tools built abroad often miss: patients and doctors switch between Hindi, English and regional languages in the same sentence. The NHA has asked for an open-source tool that handles this mix, rather than a closed product it would have to rent.

The choice also tests a model built in India. Parrotlet-A 2.5 Pro is an "open-weight" speech model, which means its trained parameters are published so others can run and inspect it. EkaCare says the model recorded the lowest error rate among the models it tested across five benchmarked languages. That is the company's own benchmark, and no independent test of it has been reported. "Lipi", described as an agentic harness, is the software layer that directs the model step by step to produce a finished note.

The Voice Scribe sits inside a wider push to put AI into Ayushman Bharat. At the Kolkata review on 5 and 6 November, the NHA also demonstrated Sandbox 2.0, an AI-assisted version of the testing environment that hospitals and software firms use to connect to ABDM, and discussed the operation of the National Health Claims Exchange (NHCX), which moves insurance claims between hospitals and insurers. Officials urged hospitals empanelled under AB PM-JAY to adopt ABDM-enabled hospital information systems to improve claims processing. The meeting also discussed BODH, a platform for benchmarking health AI models on Indian data. NHA chief executive Dr Sunil Kumar Barnwal said the next phase must move beyond expanding coverage to making sure beneficiaries get better-quality care.

Peepals calculation: officials at the Kolkata review said West Bengal generated 4.2 crore Ayushman cards and gave treatment to 2.2 lakh people in its first 80 days in AB PM-JAY. That is about 2,750 treatments a day (2,20,000 / 80 = 2,750), and treated patients equal about 0.5% of the cards issued (2.2 lakh / 420 lakh = 0.52%). Each treatment is a consultation that needs a record, which shows the scale of note-writing a working voice scribe could take over in a single state. Our calculation from the reported figures.

Short-term: The NHA and EkaCare move from selection to deployment on NHA infrastructure. Neither has given a rollout date, a list of facilities or the cost of the arrangement, so it is not yet known which hospitals will use the scribe first.

Long-term: If the tool works in crowded clinics, more consultations in public facilities could produce digital records linked to ABHA, which the ABDM needs to build complete patient histories. Accuracy in noisy wards, patient consent to recording, and how errors are caught before a doctor signs off will decide whether doctors trust it.

Who is affected: Doctors and patients in public health facilities, ABHA account holders, hospitals empanelled under AB PM-JAY, health software firms connecting to ABDM, and Indian AI developers building speech models for local languages.

Questions and Answers