top of page

Can India's Health Stack and LLMs Achieve Healthcare Independence for the Bottom 90%?

India is celebrating its 77th independence day, however, there are many other independences lined up for our people. Unfortunately, we were not the part of 1947 independence, we will be part of many others and Healthcare is one of them. And this irrational belief has a lot to do with my past failures.

I have a long list of failed Startups in the past six years, however, one among those is super interesting - and today I can laugh thinking about my irrationality and illogicality. But thanks to that failure, today we have no FOMO related to LLMs and building the fundamental model. I know the title of this essay is on the impact of India’s Health stack and LLMs on the bottom 90%. But to give you the complete context, I must navigate this essay through my story - feeling unluckiest to feeling lucky!

Let’s get started…

In late 2018, right before I departed from Maptag, I embarked on a journey to establish my own startup, despite lacking prior experience. Looking back, it's clear that I made some less-than-ideal choices, largely driven by the prevailing trends and buzzwords of the time. The year 2018 marked the initial hype around AI-based startups in India, and I distinctly recall a shift in investor interest when we presented our first deck for an AI startup. Naively, we equated receiving calls and meetings from investors with imminent funding success, even going so far as to include logos of Angel firms in our pitch deck to create an illusion of investor interest (lol). Reflecting on these missteps now brings a smile to my face, reminding me of how much I've learned and grown since those days.

Comparing my past self to who I am today reveals a stark difference. Present-day me possesses the same irrational optimism and audacious dreams as the 2018 version, but I've also acquired the skills to strategize, execute plans, solve intricate problems, and persuade others to contribute their time and effort without financial incentives. These capabilities were sorely absent back in 2018. In hindsight, I can't fathom how I could have ever conceived notions of building self-driving cars or AI developer tools without the relevant expertise or skills. Yet, those very instances of irrationality and naivety have played a pivotal role in shaping the individual I've become. So, to all the young minds out there, a touch of irrationality might not be such a bad thing after all!

People often label me as curious, but the truth is, around 2018, I was far from being well-versed in the fundamentals of startups, product development, finance, operations, programming, or anything that I'm proficient in today. Armed with inadequate skills and a somewhat misguided idea, my team and I ambitiously embarked on developing AI developer tools for enterprises. These days, I place little emphasis on others' opinions regarding the potential of our ventures. Instead, I'm driven by a commitment to first principles and a deep understanding of customer needs. It's a departure from the copycat mentality I once had.

Back when the Indian startup scene was riding the crest of the first AI wave, I fixated on the most successful AI company by funds raised. In 2018, SenseTime, a Chinese AI giant, had amassed billions of dollars in funding within a mere nine months. Inspired by this, I resolved to build India's own version of SenseTime. Unfortunately, the copycat mindset led me to mimic everything from their website to use cases, business models, and even website colors. All the while, my team members raised valid concerns about the daunting workload, yet I remained undeterred, underestimating the effort required for the initial version and glossing over the complexities of delivering a production-ready product.

At that time SenseTime was a 5000+ team member generating around half a billion dollars in revenue. And here I was dreaming of completing with them with 2 full-time and a few interns. (lol). And it is not the case that we failed to build products, the fact is we built all those APIs for text, image, video etc even generated revenues with that limited resources and time. However, even with a ready product and initial traction - we were only getting investor calls, not money.

Believing that funding equates to startup success, we dedicated ourselves to becoming investment-ready. Over the course of 18 to 19 months, we diligently crafted multiple products, diverse use-cases, and robust APIs. However, despite our efforts, we found ourselves unable to secure investor funding. Interestingly, during this same period, more than 100 AI startups secured funding right before our eyes, leaving us with a sense of being profoundly unlucky. As the saying goes, it's often hard to connect the dots when looking forward; but upon reflection, those dots reveal a different narrative, which I'll share at the end of this essay – a feeling that has shaped my perspective.

Today, as we find ourselves in a similar hype cycle, the two years of experience have become invaluable assets guiding us to make informed decisions in pursuit of the impactful and substantial company I've always envisioned.

In October of the previous year, with the public release of ChatGPT, I experienced another surge of excitement. This time, my focus shifted towards identifying a solution that could address the pressing needs of potentially billions of individuals while considering various influencing factors that could either enable or hinder their access to this transformative technology. We immersed ourselves in engaging conversations with potential users, earnestly striving to keep the core problem, users, partners, and technology in harmonious balance. Our aim was clear: deflate the cost of healthcare delivery, enhance access to quality healthcare providers, and make health insurance accessible and affordable, all within well-defined constraints.

Fortuitously, we identified the problem early on and began brainstorming solutions. One idea that emerged was leveraging India's Health Stack, should it succeed in establishing a centralized health database. While the exact form of the potential AI model remained elusive, we recognized the significant potential for it to contribute to solving one of India's most critical challenges. Curiously, we never approached LLMs (Large Language Models) as a distinct technological concept; rather, we viewed them as a tool that could aid us in addressing this vital issue.

One of the most rewarding aspects of developing such solutions is witnessing the journey from initial imagination to crafting the first prototype, demo, or MVP. When that pivotal moment arrives, and the first reaction is met with awe, it's a testament to the countless hours spent refining the concept, engaging with thousands of users and providers, sketching wireframes, considering myriad solutions, and ultimately distilling them into tangible embodiment. Throughout this process, we held a steadfast principle etched in our minds: the solution must be scalable to serve billions of people, accessible through a simple mobile phone. When the culmination of these efforts materializes, it can either lead to profound disappointment or an exultant "Wow." After enduring multiple setbacks, our most recent MVP elicited a resounding "Wow" – a gratifying testament to our perseverance and innovation.

But to explain this WoW movement, I need to explain the impact of India’s Health Stack (ABDM) on the bottom 90% of the population (It will have an impact on the entire population, however, we are building for the bottom 90% and hence only from that pov).

  1. Auto digitization and storage of Health Records

  2. Converting Health Records into an Asset for Patients

  3. Health data interoperability across Patients, Providers, and Payers

  4. Alignment of Healthcare to Payers

Auto digitization and storage of Health Records: If anyone recalls the struggles with the Wallet system - where money is added from an account to the wallet, often involving multiple UI screen switches before and during transactions - they would remember the frustrations of transaction failures amidst these UI transitions, leading to a subpar user experience. However, UPI transformed bank accounts into wallets, and thanks to the UPI architecture, transactions could be carried out without necessitating a UI switch, providing users with a seamless and efficient experience. ABHA has introduced a similar seamlessness to the process of digitizing and storing Health Records.

India struggled to have even a decent Health records database because a. 90% of the Providers are offline and b. The entire process of digitization and storage of Health Records was broken. ABHA has solved this problem, in terms of experience, this is as same as using UPI for payment. The real magic would be when providers are also on the ABDM network. The health records will get auto-attached with the user’s ABHA and will be accessible for a lifetime.

Converting Health Records into Assets for Patients: When examining the various stakeholders in the Healthcare ecosystem, encompassing Patients, Providers, and Payers, one recognizes that a substantial portion of the overall Healthcare budget is allocated to establishing a "Source of Truth," evaluating risk profiles, and validating initial cause-and-effect scenarios. Curiously, these functions all revolve around the Health Records, Lab tests, Electronic Medical Records (EMRs), Electronic Health Records (EHRs), and the like. The ability for Providers and Payers to access Patients' existing Health Records can potentially eradicate this entire expenditure. Thus, the Health Records linked to a Patient's ABHA are an asset, which Providers and Payers are willing to pay to eliminate the prevailing costs. Furthermore, the availability of historical Health Records also holds the potential to curtail Healthcare delivery expenses.

Health data interoperability across Patients, Providers, and Payers: Back in February 2023, as I penned the first essay regarding the launch of Jile Health, my curiosity was piqued by the perplexing scenario of the United States' Healthcare expenditure reaching a staggering million dollars per capita, despite a remarkable 95% digital penetration, an 85% adoption rate of Electronic Health Records (EHRs), widespread access to cutting-edge technologies, approximately 50% telemedicine penetration, and an impressive 85% healthcare insurance coverage. Our inquiry led us to a crucial realization: despite the prevalence of EMR adoption by healthcare providers, a substantial portion of this data remained isolated within individual provider systems, rather than being shared among multiple stakeholders to amplify its value and concurrently mitigate the expenses associated with Healthcare delivery.

Thanks to the ingenious architecture of India's Health Stack, ABDM (Ayushman Bharat Digital Mission). This intricate design has effectively addressed the challenge at hand, largely due to the seamless interoperability of ABHA. By virtue of being linked with ABHA, Patients' Health records can be effortlessly shared with any participating providers and payers across the ABDM network. This groundbreaking feature stands as a key factor influencing our confidence that the per capita expenditure on Quality Healthcare in India will remain well under INR 10,000, marking a stark contrast to the figures observed in the United States.

Alignment of Healthcare to Payers: The pivotal factor underpinning the success of companies like United Health Group (UHG) is the strategic alignment of Healthcare with payers. With the comprehensive solutions outlined earlier, achieving this alignment becomes notably more feasible, enabling the creation of personalized, incentive-driven Health Insurance premiums.

Let’s understand through an example: Over the past five decades, India's Health Insurance landscape has witnessed minimal transformation, primarily due to various technological constraints. Consequently, Health Insurance offerings from different providers tend to exhibit a remarkable similarity.

The determination of Health Insurance premiums revolves around several factors, encompassing Agent Commission, Operational Costs, Distribution expenses, Claims, Reinsurer commissions, and profits. While I once presumed that reinsurer commissions amounted to around 5%, the actual figure is quite different. In India, the average reinsurer commission stands at 20%, as per IRDAI (Insurance Regulatory and Development Authority of India). It's worth noting that manufacturers are compelled to reinsure the risks they underwrite when offering health insurance policies. Examining prominent reinsurers like GenRe reveals that these commissions hinge on the risk profile, encompassing factors such as Critical Illness, Long-Term Care, Longevity, and overall Health. Introducing a straightforward risk assessment tool has the potential to reduce premium costs by a significant margin, approximately 10% to 15%, without necessitating major modifications. However, constructing such tools remains an elusive task unless the alignment dilemma between Healthcare and Payers is resolved.

Once this alignment challenge is effectively addressed, the focus shifts to leveraging Patients' Health records to substantially curtail Healthcare delivery expenses. This process commences with the development of advanced AI models capable of deciphering Health Vitals extracted from Patients' Health Records. By enabling affordable Healthcare delivery through the creation of hyper-localized, individualized solutions for India's bottom 1.2 billion population, the potential for reducing Healthcare costs on a significant scale becomes a tangible reality.

We already have initial success stories with Patients with multiple diseases - Chronic, Mental illness, Muscle problems etc.

You won’t believe for the initial 8 months, we were doing everything keeping the problem statement in our mind - “Making Quality Healthcare affordable for the bottom of 90% (1.2 billion Indians)” - and we, in fact, struggled to explain what we were building. For the next 2 months, we spotted potential steps to solve the above problem, and in the last two months, we have been building a few of the potential use cases that will help us solve this problem once and for all.

We ourselves have watched these use cases uncountable times because we are still processing how did we come to the solutions. And thanks to Health Stack and LLMs independence of quality Healthcare for the bottom 90% is inevitable...

We need to come back to the last part of the story: feeling unluckiest to feeling lucky! So, do I feel unluckiest anymore looking backwards? The answer is a NO because when I am looking around, there is not even one of those funded startups existing or becoming a large startup. Most of them are Zombie startups or just one or two of them are 10+ million ARR-size startups. I am not signalling my happiness about these Startups failing to make it big. You must have found me on the front to celebrate their success.

I am not making this up however, I never dreamt of building a small startup. There was a sense that whatever we will build it must impact billions. I don’t know how did we survive? Or how did I survive for around 2 years without funding in the world’s one of the costliest cities - Bangalore - with negative skills, irrational thinking, and full of naivety? But I survived. And there is a saying what doesn’t kill you makes you stronger. Today I can write this will full confidence - I was not the unluckiest in the first AI hype cycle by not getting funded. So, in that sense, luck has been kind to me but not favourable yet - I am yet to get lucky. :)

Before I wrap up this essay, Happy Independence Day everyone. Just remember we have many independences lined up for our people! Thanks for reading, If you like this essay, please share this with your network. I shall see you all the next week

Recent Posts

See All


bottom of page