EkaCare Selected by NHA for AI Voice Scribe Solution Under ABDM

- EkaCare has been selected by NHA as a technology partner for its Ambient AI-enabled Voice Scribe under ABDM.
- The solution supports more than 15 Indian languages and generates structured clinical notes in FHIR format.
- The platform uses EkaCare’s Parrotlet-A 2.5 Pro speech model and open-source Lipi agentic scribe.
Bengaluru-based health AI startup EkaCare, operated by Orbi Health Private Limited, has been selected by the National Health Authority (NHA) as a technology partner for its Ambient AI-enabled Voice Scribe solution under the Ayushman Bharat Digital Mission (ABDM).
Announced on November 11, 2026, the selection followed NHA’s Expression of Interest (EOI) process for open-source ambient AI-enabled voice-to-text solutions for digital health record creation. The process included a demonstration before NHA’s Evaluation Committee. The solution will run on NHA infrastructure powered by Amazon Web Services (AWS).
EkaCare also demonstrated the technology during NHA’s National Review Meeting with States and Union Territories in Kolkata on November 5 and 6. The company’s Founder and CEO, Vikalp Sahni, and VP of AI Engineering, Neha Jagadeesh, presented its speech model and open-source clinical documentation technology.
How EkaCare’s AI Voice Scribe Works
The Ambient AI Voice Scribe is designed to listen to doctor-patient consultations and convert conversations into structured clinical notes. It supports more than 15 Indian languages, including multilingual conversations in which participants switch between languages.
The generated notes capture patient complaints, medical history, medications, vital signs, laboratory findings, diagnosis and treatment plans. They follow speciality-specific templates and use the Fast Healthcare Interoperability Resources (FHIR) format, which supports the exchange of structured health information across digital healthcare systems.
Doctors retain control over the final records. Each AI-generated note must be reviewed, edited if required and approved by a verified doctor before it is saved and linked to the patient’s Ayushman Bharat Health Account (ABHA). Doctors accessing the standalone platform will authenticate through the Healthcare Professionals Registry (HPR).
The solution can be integrated into existing Hospital Management Information Systems (HMIS) through an embedded mode. Hospitals can also access it through a standalone interface or APIs based on the open MedScribe Alliance standard for medical scribes.
EkaCare says the system is designed to operate entirely within NHA’s secure, air-gapped environment, with the infrastructure hosted on AWS. This approach is intended to allow public healthcare facilities to retain control over their health data.
Parrotlet-A 2.5 Pro and Lipi Power the Platform
EkaCare’s Voice Scribe is built around two technologies developed by the company: Parrotlet-A 2.5 Pro, an open-weight speech model designed for Indian healthcare conversations, and Lipi, an open-source agentic scribe previously known as Vaarta.
According to EkaCare, Parrotlet-A 2.5 Pro recorded the lowest error rate among the models evaluated across five benchmarked languages in a dataset of 6,550 real Indian medical consultations. Its model weights are publicly available through AIKosh and Hugging Face.
Lipi converts conversations into structured records and can also be used for applications beyond clinical documentation, including meetings and other conversational workflows. EkaCare publicly open-sourced Lipi on November 11, 2026.
The NHA selection builds on the company’s existing AI documentation product, EkaScribe, which EkaCare says is used by more than 1,000 doctors in private practice.
Commenting on the development, Bhaskar Chaudhary, Chief Product Officer at EkaCare, said the technology is intended to help doctors focus on patients rather than documentation. He also highlighted the development of Parrotlet-A and Lipi around India’s diverse languages, accents and speech patterns.



