Bengaluru: Eka Care has introduced Parrotlet-a v2, India’s first clinical-grade, real-time AI model for medical transcription, designed specifically for hospitals and outpatient departments. Unlike general AI systems, Parrotlet-a v2 is built to handle Indian clinical language, including Hindi and Indian English, regional accents, overlapping conversations, and local drug names. Doctors can now convert patient interactions into structured medical notes almost instantly, reducing the heavy documentation workload and allowing more focus on patient care.

“India’s hospitals are busy, multilingual, and noisy, which makes global AI systems unreliable here. Parrotlet-a v2 is built for Indian healthcare, delivering fast and accurate medical transcription during real consultations,” said Vikalp Sahni, Founder & CEO, Eka Care.

In tests on Indian clinical datasets, Parrotlet-a v2 achieved 93% medical keyword accuracy, outperformed models like Gemini 3 Pro, Gemini 3 Flash, and Saaras V3, and maintained a very low Phantom Speech Rate of 3%, meaning it rarely mistakes background noise for clinical information. It works up to 30 times faster than large AI models, providing doctors with accurate notes in seconds rather than waiting for lengthy processing.

The model powers EkaScribe, Eka Care’s AI scribe platform, now used by over 3,000 doctors, bridging efficiency with intelligence while staying cost-effective for large-scale use. With this launch, Eka Care aims to reduce administrative burdens, support healthcare digitisation across India, and enable doctors to focus on what matters most—their patients.