Artificial Intelligence: The Next Layer of Healthcare

By Yash Gorkhoo and Tanvi Dubey

Healthcare has not lacked technology. It has resisted reorganisation. Every major wave — from EMRs to telemedicine to wearables — improved specific functions, but required the system to change around them. Adoption was slow because the burden sat with the clinician.

Each drove incremental change. AI is different- not because it improves specific functions, but because it begins to sit inside decisions. That is a different kind of entry point.

The System Underneath

Healthcare in India spans a familiar set of layers- diagnostics, delivery, devices, pharma, services, and preventive care. On paper, it looks complete. In practice, it behaves very differently.

Healthcare remains a doctor-led system. Clinical authority sits with the physician, and decision-making flows through them. Hospitals, despite their capital intensity, are largely infrastructure plays- they enable care but rarely control it. And unlike most developed markets, the patient is still the primary payer. Out-of-pocket expenditure dominates. Insurance is growing, but it has not redefined the system. Despite expansion of schemes like Ayushman Bharat: Insurance and financing scheme coverage reached 46% in rural areas and 32% in urban areas in 2025, Nearly 400 million Indians remain uninsured and are financially vulnerable to medical emergencies. This shapes pricing, access, and behaviour in ways that are fundamentally different from insurance-led markets.

India’s healthcare system is often described as efficient- and in some ways, it is. It delivers complex care at a fraction of global costs. Cardiac surgeries, oncology, orthopaedics- high-quality treatment exists at a fraction of global costs. Coronary Artery Bypass Graft (CABG) USA: $150,000–$200,000. India: $7,000–$12,500 all-inclusive (including ICU stay, follow-ups, and initial companion expenses), Single Chemotherapy Cycle :USA: $5,000–$10,000 per cycle. India: ₹15,000–₹2 lakh (approximately $180–$2,400) per cycle for identical molecule

But that efficiency is the product of constraint, not design.

We have seen how previous technology waves asked healthcare to reorganise around them. EMRs demanded new workflows, heavy IT investment, retraining, and regulatory alignment. Adoption was slow because the burden of change sat with the clinician- and the trade-off was rarely worth it.

A system under stress is also a system with pent-up demand for things that actually work. The gaps in Indian healthcare- access, productivity, data- are not marginal inefficiencies. They are structural.

This is where AI behaves differently. It inserts itself at points of maximum friction– improving outcomes without demanding change upfront.

Where the System Breaks: Access, Productivity, Data

Three structural failures define Indian healthcare: access, productivity, and data.

The first is access. There are not enough doctors, and not in the right places. A system where a specialist visit requires a six-hour journey is not accessible in any meaningful sense. Technology that extends diagnostic capacity- AI-led screening, triage systems, remote monitoring- does not replace the specialist. It expands their reach.

Redcliffe Labs — one of India’s fastest-growing diagnostics networks and a Chiratae portfolio company — is a strong example of this. Through an omnichannel model combining home sample collection, digital results, and an expanding physical footprint, Redcliffe has made quality diagnostics accessible across Tier 2 and Tier 3 towns — places where a lab visit was previously not a given. Redcliffe Labs recently introduced an artificial intelligence powered facial scan that provides a fast, contactless wellness check using any smartphone or camera-enabled device. The tool is able to determine the most important parameters, including the heart rate, the heart rate variability, the breathing rate, and the blood pressure by comparing the subtle colour variations in the skin due to blood flow

The second is productivity. The clinicians who do exist are overloaded. Documentation, coding, discharge summaries, insurance workflows consume hours that should go to patients. At Indian patient volumes, this problem is more acute than anywhere else. Every hour reclaimed from administrative overhead is an hour returned to care.

Healthplix – an EMR platform built for Indian physicians and a Chiratae portfolio company – is a good example of what narrow workflow automation looks like in practice. It enables doctors to generate a prescription in under 30 seconds, with specialty-specific modules that fit naturally into clinical workflows. The result is less time on documentation, better patient retention, and an online consultation layer that extends the doctor’s reach – all without asking them to change how they practice.

The third is data. Healthcare generates continuous signal-labs, imaging, vitals, prescriptions, clinical notes. The sector is, in a very real sense, data-rich but intelligence-poor. Almost none of it is connected, and almost none of it compounds. Records do not travel. Trends remain invisible. Population-level insight stays largely untapped. ABDM begins to change this, it creates the rails for interoperability. But the infrastructure is early, and adoption is uneven.

When this data is structured and made actionable, it compounds. Enabling more personalised treatment protocols, sharper preventive interventions, and over time, measurably better clinical outcomes.

Where AI Enters First

The first meaningful impact of AI in healthcare is not in diagnosis or treatment, at least not first. It is in the layers that surround care- administration, workflows, and decision support.

To understand the scale of this opportunity, the US is instructive. Healthcare spend there is approaching $5 trillion. Of that, a significant portion flows through non-clinical workflows- revenue cycle and billing alone represent hundreds of billions in recurring spend, much of it still running on manual processes, fragmented systems, and paper records. The administrative layer is enormous, and it is inefficient by design. At the non-clinical layer, AI is already being deployed across revenue cycle management, documentation, scheduling, and claims. These are large cost pools with clear inefficiencies. Regulatory friction is low, and ROI is measurable. The result is predictable: adoption is faster.

US Healthcare Spend	Amount	% of Total
CLINICAL SPENDING (Direct Patient Care)	$3.54T	67%
ESTIMATED NON-CLINICAL (Admin/Overhead)	$1.2 - $1.6T	23-30%
Other (Investment, Research, Public Health)	$0.5T	9-10%

Source: Centers for Medicare & Medicaid Services (CMS) – National Health Expenditure Accounts 2024

The opportunity in non-clinical workflows is significant – and nowhere is this more visible than in the US, where administrative overhead consumes a disproportionate share of healthcare spend. A new generation of companies is attacking this from specific, well-defined wedges. Rapid Claims and Arintra are going after revenue cycle management — one of the most fragmented and manual parts of the system. Coral is focused on administrative workflow automation. And Abridge has built a compelling position starting with AI scribing – reducing the documentation burden on physicians – and is now expanding into adjacent workflows. What’s instructive about Abridge is the wedge strategy: earn trust by solving one painful problem well, then expand from that position into the broader workflow layer.

What looks incremental at first- reducing manual effort, improving throughput, tightening workflows- compounds quickly. These systems begin to shape how information flows, how decisions are triggered, and how outcomes are tracked. This is not glamorous but definitely foundational. We are closely tracking this space and excited to see new business models that will emerge from this segment.

The Clinical Shift

On the clinical side, the potential is larger- diagnostics, decision support, treatment pathways, drug discovery. Progress here is slower. The constraints are real: regulation, trust, liability. The feedback loops are longer and the tolerance for error is lower.

Which is why the sequence matters. Operational intelligence precedes clinical intelligence. Efficiency builds trust and trust enables adoption. And over time, the two converge into a single system.

But this sequence does not play out the same way everywhere. It is shaped by the structure of the system it enters.

India vs the US: Same Shift, Different Starting Point

AI does not arrive in a vacuum. It enters the system that already exists. In the US, healthcare is insurance-led. Technology spend is high. Digitisation is mature. AI adoption is, in many ways, a displacement story – automating an already large administrative layer. India is fundamentally different. It is consumer-led, cost-sensitive, and unevenly digitised.

And the numbers reflect this gap. India’s total healthcare spend is around $218 billion- a fraction of the US market ($5 Trillion) – and digital healthcare accounts for barely 1-2% of that today. But that number is projected to reach 5-6% by 2030, which means the digital layer is set to triple in relative terms, even as the overall market grows. The opportunity is not in displacing a mature system. It is in building the digital layer from the ground up.

Which is why AI in India is not replacing doctors. It is extending them. In a system where doctor bandwidth is constrained and access is inconsistent, AI becomes a force multiplier- it allows the system to do more with the same core resource.

This changes the adoption equation entirely. In India, the more urgent question is access- how quickly capability can reach populations that currently have none.

The Role of the Doctor

At the centre of this shift is a question the industry is actively navigating: how will the doctor’s role transform in an AI-enabled system?

Historically, general practitioners have performed two primary functions: pattern recognition and knowledge application. Medical memory—protocols, clinical literature, diagnostic references – has increasingly been externalized through structured systems and evidence repositories. Pattern recognition, by contrast, has remained where clinical expertise traditionally concentrates. And it is precisely here that AI shows significant capability.

This creates a structural tension:The technology demonstrates capability faster than regulatory frameworks, clinical governance structures, and professional consensus can accommodate it. This gap – between what becomes possible and what the system permits—shapes the actual timeline for clinical adoption.

Key implications worth considering:

Role redefinition, not elimination: Doctors may increasingly focus on judgment, context integration, patient relationships, and clinical decision oversight rather than pattern detection alone
Geographic variance: Markets facing acute healthcare access pressure—whether rural regions or developing economies—may experience accelerated adoption timelines, as the cost-benefit calculus shifts differently than in resource-rich settings
Capability asymmetry: The ability to deploy AI diagnostics typically outpaces institutional readiness to integrate them into workflows, reimbursement, and liability frameworks
Human-AI teaming as baseline: Rather than replacement, the emerging model centers on collaborative workflows where AI handles specific tasks and clinicians manage judgment, exceptions, and patient communication

The Consumer Shift

AI is not only reshaping clinical systems. It is changing behaviour. But the shift is not limited to the doctor.

Healthcare is moving from episodic to continuous, from reactive to managed, from institution-centric to increasingly consumer-influenced.

In India, this shift is amplified by the fact that the consumer already pays. Prevention is becoming a category people spend on- tracking, nutrition, fitness, early intervention. The model is evolving toward subscription, behaviour change, and ongoing engagement.

Preventive health spend pool is also increasing due to following reasons:

Diabetes Epidemic – Accelerating

India’s diabetes prevalence rate is projected to increase steadily from 6150.19 per 100,000 in 2022, reaching 6960.33 per 100,000 by 2025, and rising to 8585.45 by 2031. More directly: Among adults aged 45 years and older, 19.8% have diabetes, with 60.1% aware of their condition. The National Programme for Prevention of NCDs (2023–30) commits to placing 75 million patients with diabetes or hypertension on standard care by 2025.

Elderly Population – Exponential Growth

The number of elderly in India is projected to reach 158.7 million in 2025, up from 150 million in prior years. The share of the elderly population aged 60+ is expected to rise from 8.6% in 2011 to 10.1% in 2021 and 13.1% in 2031.

Care Access Crisis

Two-thirds of India’s elderly live in rural areas where healthcare access, pension awareness, and infrastructure are significantly weaker. India has fewer than 1,000 geriatricians for over 150 million elderly people

A new category of D2C chronic care platforms has emerged – BeatO and Sugar.Fit are among the more established players in diabetes management, combining connected devices, coaching, and care protocols into a continuous model. The weight management space is seeing similar momentum, with platforms like ElevateNow building programmes that sit at the intersection of clinical and lifestyle intervention – increasingly relevant as GLP-1 drugs begin to reshape what’s possible. And companies like Prana Health are going broader, building infrastructure for multi-condition chronic care management. These are not health apps — they are longitudinal care relationships, enabled by technology.

GLP-1 drugs accelerate another shift — the convergence of clinical intervention and lifestyle outcomes. The line between healthcare and wellness is beginning to blur.

Where We’re Paying Attention

The infrastructure is being built. The question now is who builds on top of it- and what they build first.

The companies that will define Indian healthcare AI are not the ones trying to solve everything. They are the ones that have picked a specific point of friction- a workflow, a data gap, an access problem- and gone deep on it. Narrow entry, compounding moat.

At Chiratae, we are particularly excited about a few spaces — AI in Health solving for the US market, where non-clinical workflow automation represents an immediate and large opportunity across delivery, pharma, and payors; Consumer Health in India, across preventive wellness and outcome-driven chronic care; and Single Specialty plays building depth through focused, technology-first execution. These are not the boundaries of where we look — they are where we see the most immediate signal today.

That view is shaped by over $130 million deployed across the stack — from diagnostics (Redcliffe) and health tech (Healthplix, HexaHealth) to consumer health (Cult.fit, HealthifyMe), single specialty (Dezy), med-devices (Aether Biomed, SigTuple), biotech (Zumitor Biologics), and insurance (ClaimBuddy).

The common thread is not the technology. It is where the product sits- between signal and decision, between patient and system, between data and action. That position, once earned, is hard to displace.

India’s digital healthcare market is early enough that foundational positions are still being earned- and large enough that the companies that earn them will matter at scale. The window is open. We are backing the builders who see the system as it is, not as it should be and are building for that reality.

If you are building at the intersection of AI and healthcare, whether in the US market or India, we would love to hear from you. Reach out to us at yash@chiratae.com yash@chiratae.com

Sources:

Chiratae Sonic DeepTech Program Applications extended till 16th Feb 2026 - Apply Now