On August 7, the National Health Authority announced that the ABHA-based Scan and Register service had crossed 25 crore OPD registrations, or 250 million registrations. It is now operational across 30,800 healthcare facilities in 756 districts, with nearly four lakh, or 400,000, registrations taking place every day.
For anyone who has spent time inside the Indian healthcare system, this is not a small change. A patient can walk into a participating hospital, scan a QR code, share their ABHA profile with consent, and register without repeatedly filling out the same information. What used to be a largely paper-based interaction is increasingly becoming part of a common digital infrastructure.
But registration is still only the beginning. Once the healthcare system knows who the patient is, the harder question is whether it actually knows the patient.
Knowing who the patient is, is not the same as knowing the patient
A hospital may now be able to identify a patient digitally when they arrive, but identity is only the first layer of clinical context. The doctor still needs to know whether the patient's creatinine has been gradually increasing over four years, what a CT scan at another hospital showed six months ago, which medicines another specialist prescribed, and whether the symptom being described today has appeared in earlier consultations.
Most of this information may already exist somewhere, but it rarely exists together. One hospital has the discharge summary, a diagnostic laboratory has the blood results, an imaging centre has the scan, and a specialist has the prescription. The patient may also have PDFs on WhatsApp, paper reports at home, and records sitting across different hospital portals.
We have spent a lot of time in healthcare talking about digitising these records, and that work is necessary. But digitisation is not the same thing as continuity, and a hundred disconnected digital records are still a hundred disconnected records.
Healthcare does not happen in visits
A hospital registration is an event, but a patient's health is not. Health changes over months, years and often decades, and the value of medical information often comes from understanding what changed, what preceded it, and what happened afterwards.
A single HbA1c result tells a doctor something, while five HbA1c results over three years tell a different story. A CT scan has more meaning when an earlier scan is available for comparison, and a prescription becomes more useful when it can be seen alongside the diagnosis, laboratory values and subsequent response to treatment.
The same applies to clinical encounters. Something a patient tells a doctor today may appear insignificant in isolation, but become important when it is the third time that symptom has appeared over eighteen months. This is why continuity becomes more important as the volume of digital health data grows.
Moving data is not enough
The Ayushman Bharat Digital Mission is creating the rails through which health information can be identified and exchanged with patient consent. That is foundational, but once those rails exist, the more difficult question is what happens to the information moving across them.
Does a laboratory report remain a PDF, or do the values inside it become part of the patient's history? Can a blood result from one laboratory be compared with a result from another? Does an imaging study remain an isolated scan, or can it be understood alongside earlier imaging and the clinical events around it?
For this to work, the individual pieces have to remain connected to the patient and to each other. Reports need to be structured, laboratory values need to be normalised, and imaging, prescriptions, vitals and clinical encounters need chronology and provenance. The objective is not to build a larger digital filing cabinet, but to preserve the patient's story.
This is closely related to a problem we have written about before: the real battle in ABDM interoperability is data quality, not simply APIs. Moving information between systems is important, but the information also has to remain usable once it arrives.
This is the problem we are working on at Aether
Aether started with a simple problem: medical reports contain enormous amounts of useful information, but most of it remains trapped inside documents. As we built further, it became clear that extracting the report was only the first step and that the larger problem was the longitudinal record underneath it.
A patient is not their latest report, prescription or consultation. Their health is the accumulation of thousands of clinical events, measurements and decisions across time. What we are building now is a longitudinal intelligence layer around that patient.
Reports, imaging, prescriptions, vitals and clinical encounters are structured and connected into a continuously evolving health graph. New information does not simply get stored. It changes what the system knows about the patient, how current findings relate to earlier ones, and what may deserve attention next.
This is also where agentic AI becomes interesting in healthcare. Instead of waiting for a doctor or patient to ask a model a question, an agentic system can work continuously across the longitudinal record. A new laboratory report can be compared against years of previous results, a medication can be understood against diagnoses and earlier prescriptions, and a new imaging finding can be connected to prior scans and clinical events.
Changes, contradictions and emerging patterns can then be surfaced as the patient's record evolves. The health graph effectively becomes memory for these agents, giving them persistent context rather than forcing them to reason from whatever documents happen to be available at that moment.
That is very different from putting a chatbot on top of a collection of medical PDFs. The intelligence comes not only from the model, but from having a persistent, structured understanding of the patient underneath it. As medical models become more capable, this distinction will matter even more because the harder problem will be giving those models complete, reliable and longitudinal patient context on which to reason.
We have made a similar argument in Everyone Wants to Build the AI Doctor. The model may become increasingly commoditised, but the depth and quality of patient memory underneath it will remain difficult to build. That is the layer we are building Aether around.
What comes after 250 million registrations
The 250 million Scan and Register milestone matters because the front door of Indian healthcare is becoming digital. The more interesting challenge begins after the patient walks through it, when every report, scan, prescription, diagnosis, vital and consultation has to become part of a history that follows the patient and becomes more useful with time.
That is when digitisation starts becoming something more valuable than a faster transaction. It becomes continuity, and eventually the memory layer on which better clinical decisions and more capable healthcare AI can be built.