HealthPlix scales AI-powered clinical documentation across Indian languages with Sarvam
"AI-powered clinical documentation is reshaping how doctors engage with patients, moving the EMR from a system of record into a real-time partner inside the consultation room. As doctors increasingly expect technology to understand the way they actually speak across India's many languages, dialects, and clinical contexts, voice-led AI is emerging as a natural way to capture consultations accurately, without taking attention away from the patient.
Our partnership with Sarvam has enabled us to deliver clinical documentation that genuinely understands Indian languages, accents, and the way doctors and patients converse in a real consultation.
By embedding multilingual, context-aware transcription into HALO, we are helping doctors save time, improve patient experience, and deliver better outcomes at scale."
Chaitanya Raju
Executive Director and Chief Product Officer, HealthPlix, HealthPlix
Background
HealthPlix is India's largest AI-powered EMR, helping doctors deliver better patient outcomes. Used by more than 14,000 doctors across the country, the platform processes approximately 1.5 lakh OPD consultations every day, across multiple clinical specialties.
As the organisation scaled its footprint and clinical reach, the need to make the EMR genuinely usable inside a real consultation became central to its product direction. To support this, HealthPlix began embedding AI-led capabilities across the consultation workflow, with a focus on voice-driven documentation that could capture conversations naturally, in the languages doctors and patients actually use, without slowing down the clinical encounter. The primary user of these capabilities is the doctor, with medical scribes positioned as a potential extension of the same workflow over time.
The Problem
HealthPlix's expansion across specialties and geographies amplified the need to support a wide range of Indian languages, regional accents, and clinical workflows. Doctors increasingly expected the EMR to keep pace with how they spoke in the room, including English-Hindi and English-regional language code-switching, brand-name medications, and clinical shorthand used in everyday consultation.
Global ASR systems struggled with this context. Accents tripped them up, Indian drug brand names came out mangled, and the way doctors actually used clinical terminology in everyday consultation was outside what those models had been trained on. The downstream cost was significant: if the raw transcript was not accurate enough, doctors had to manually edit everything, which defeated the point of the tool. For voice-led documentation to be worth using, transcripts had to be reliable enough for doctors to review, without having to rewrite them.
The priority was to establish a reliable, multilingual speech-to-text foundation that could work consistently across Indian languages, handle the variability of real OPD environments, and deliver transcripts accurate enough to drive structured medical records with minimal manual editing. The organisation focused on building a voice-powered documentation layer that could:
- Transcribe consultations accurately across multiple Indian languages and regional dialects.
- Handle code-switching, overlapping speakers, and varied audio conditions in clinical settings.
- Recognise Indian drug brand names and clinical terminology in everyday spoken form.
- Support consistent performance across specialties, including long, counselling-heavy consultations.
- Reduce documentation time so doctors could focus on the patient.
Why Sarvam
HealthPlix evaluated multiple frontier AI models against real consultation audio before settling on a partner. Sarvam consistently outperformed the alternatives on the dimensions that mattered most in an Indian OPD: prescription accuracy, interpretation of regional dialects, and the ability to follow code-switching and clinical terminology in context.
Sarvam’s biggest differentiator was accuracy in the Indian context. Sarvam's models understood how Indian doctors and patients actually speak, rather than how a sanitised English transcript assumes they speak. That contextual fit, combined with strong handling of Indian brand-name medications and reliable performance across specialties and audio conditions, made Sarvam the natural foundation for HALO.
The Solution
HealthPlix deployed Sarvam STT Streaming v2.5 within HALO, its AI-powered clinical documentation product, to power voice-led documentation across its doctor base.
Within HALO, Sarvam's speech-to-text transcribes the consultation in real time, and the resulting text is structured into the sections that make up a complete clinical record, including vitals, complaints, diagnosis, medication, advice, and treatment plan. Sarvam STT integrates into HealthPlix's stack as a standalone API call within the HALO workflow. Audio capture is configured to suit each clinic's setting, and Sarvam has performed reliably across the range of conditions encountered in a busy practice, from background noise to multiple speakers in the same consultation.
To illustrate, consider a typical OPD interaction. To make sure patients clearly understand what they are being told, doctors often set aside clinical terminology and explain things in everyday language. A doctor in a diabetology practice might say "your sugar is on the higher side" to put the patient at ease, while intending to record "Type 2 Diabetes Mellitus" in the prescription. Global ASR systems would transcribe this literally, leaving the doctor to manually correct the record after the consultation. With Sarvam STT inside HALO, the same exchange is captured in the doctor's spoken form, while the structured record reflects the clinical terminology accurately. The doctor finishes the consultation with a record that needs review, not rewriting. The same applies to Indian brand-name medications and code-switching across languages, where Sarvam consistently captures the spoken phrase and maps it to the right clinical entry, across both short consultations and longer counselling-heavy ones.
HALO is currently live across all specialties HealthPlix serves, including disciplines where consultations are long and involve detailed patient counselling. HALO is being rolled out progressively across the doctor base, with more than 25,000 consultations completed through it to date, and doctors retain full review and oversight of every record before it is finalised.
Across the doctor base, the feedback has been consistent. With Sarvam powering its underlying transcription, HALO has, in the words of many doctors, become their AI assistant during consultations. The time saved, particularly the five minutes typically spent typing prescriptions, is the most frequently cited benefit. Doctors also feel that the enriched accuracy of HALO creates a better patient experience, allowing them to maintain eye contact, listen more attentively, and focus fully on the conversation instead of being occupied with documentation.
From an engineering standpoint, working with Sarvam has been straightforward to integrate into HealthPlix's existing stack. The APIs are well-structured and optimised for real-world Indian language use cases, which reduced the customisation and fine-tuning effort required to reach production-grade accuracy. Sarvam's responsiveness and the flexibility to support healthcare-specific workflows have allowed HealthPlix to maintain both the accuracy and latency standards required in a clinical environment. For a health-tech organisation handling sensitive patient information, Sarvam's alignment with the compliance and security standards HealthPlix operates under supports the company's commitment to protecting patient data.
The Impact
- Accuracy: 97%+ accuracy on prescriptions generated through HALO, enabling doctors to review rather than rewrite.
- Time saved: Approximately 5 minutes saved per consultation, recovering the time doctors previously spent typing prescriptions.
- Adoption: More than 25,000 consultations completed on HALO to date.
- Language coverage: Noticeable improvement in transcription quality for Indian languages beyond English and Hindi, including handling of regional dialects and code-switching.
- Clinical experience: Doctors report being able to maintain eye contact with patients through the consultation, instead of dividing attention between the patient and a keyboard.
- Operational reliability: Consistent performance across specialties and clinical settings, including long counselling-heavy consultations and high-volume practices.
- Latency and cost: Sarvam STT has contributed to improvements in transcription latency and to cost efficiencies relative to alternatives previously evaluated, supporting HALO's ability to operate at scale in a real-time clinical setting.
Looking Ahead
Over the next 12 months, HealthPlix expects its use of Sarvam within HALO to deepen, with the largest growth coming from expanded language and dialect coverage, followed by higher per-doctor consultation volume, broader specialty adoption, and longer-term exploration of additional voice-led use cases such as real-time scribing and TTS read-back. HealthPlix continues to evaluate additional API capabilities and clinical use cases beyond those publicly shared, with further details kept confidential given the competitive nature of the space.
The collaboration between HealthPlix and Sarvam reflects a shared commitment to building AI that is genuinely fit for the Indian clinical context. By combining HealthPlix's deep understanding of doctor workflows with Sarvam's strength in Indian-language speech recognition, HALO is helping doctors document consultations more accurately, save valuable time, and deliver better patient experiences across the country. As HealthPlix continues to expand its footprint and Sarvam continues to advance its multilingual AI capabilities, the partnership is set to play an increasingly meaningful role in shaping the future of AI-led healthcare in India.