How HealthPlix turns doctor consultations into medical records with Sarvam

Doctors in India still spend valuable time creating medical records manually. HealthPlix built HALO with Sarvam’s Speech to Text Model to turn doctor-patient conversations into medical records at scale.

In this session, Sehaj Virk, Head of Model APIs at Sarvam, sits down with Chaitanya Raju, CPO at HealthPlix, for a behind-the-scenes look at how HealthPlix embedded Sarvam’s speech recognition into HALO to power real-time medical records across languages, specialties, and live doctor-patient conversations.

What we’ll cover:

• Why voice AI is a hard problem inside Indian OPDs, and where global ASR systems fall short
• How HALO handles code-switching, regional accents, clinical shorthand, and Indian brand-name medications in real time
• The accuracy and latency bar a real consultation demands, including 97%+ prescription accuracy
• What adoption looks like in practice - 50,000+ consultations completed, with doctors saving around 5 minutes per consultation

For anyone building speech recognition, healthcare AI, or voice-led workflows in India, this is a real look at what production actually takes.