For a decade, my AI helped doctors catch tuberculosis, blindness and disease.
Now that same machine watches over your shopping cart.

It started with a bottle of “almond” hair oil. I turned it over and read the fine print — over 90% mineral oil, barely a trace of almond. The front showed a handful of almonds and a promise. The back told the truth, quietly, in letters so small you were never meant to read them. Something broke in me that day. Not sadness — anger. How many parents had bought this for their children, trusting the picture on the front?

I am Balendra Kumar Singh, and I did not come to this from marketing. I grew up in a village, where the answer to most ailments grew in the soil around us — where people understood natural things long before a label tried to sell them back to us. I studied Physics, Chemistry and Mathematics, which taught me to actually read a chemical composition instead of fearing it; computing did the rest. Then I spent more than twenty years building software — a developer and team lead across nearly every technology there is — and for more than ten of those years, the work that shaped me was AI for healthcare. At SevaMob — an AI-assisted primary-care platform recognised by Gartner as a 2019 Cool Vendor in Healthcare AI — I built a suite of proprietary clinical AI — deep learning, machine learning and computer vision, engineered to run rapidly, offline, in low-resource settings and be used with minimal training — to put expert judgement in the hands of frontline technicians and doctors: Vision AI to screen for cataract, diabetic retinopathy and glaucoma; Sputum AI to detect tuberculosis; Blood and X-ray point-of-care screening; lab-report analysis; Symptom AI to read a patient’s symptoms and suggest likely conditions; a health score with personalised recommendations; and Diet AI to tailor nutrition to a person’s allergies, age, gender, region and goals — all exposed through a developer API that other health apps build on. None of it was built for consumers — every model existed to assist a medical professional. For a decade, my job was to teach machines to look at the human body and tell the truth about it — even when the truth was hard, even when a life depended on getting it right.

The problem that changed everything was personal. My wife and I had gone grey in our twenties. We tried everything — nothing worked, and worse, nothing was honest. Then, around 2018, I saw the same early signs in my daughter, and a vague worry hardened into a father’s quiet, fierce determination. So the engineer who’d spent years reading medical data started reading something else: labels. The chemical names. The E-numbers. The FSSAI codes nobody explains. Drawing on the natural wisdom I grew up with, I began making honest products of my own — the work that, in 2025, became Botani Bestie: an Ayurvedic hair- and skin-care brand built on one rule — real botanicals at honest concentrations, no sulphates, parabens or silicones, and nothing to hide on the back of the pack.

Then it stopped being about hair and skin — and became about a life. A close friend’s father, a heart patient living with high blood pressure, did everything a careful man is told to do: cut back on salt and fat, and reach for the “healthy”, “light”, “heart-friendly” packaged foods instead. He trusted the promises on the front of the pack. What he had no way to read was the back — the sodium, the refined palm oil, the additives a reassuring label quietly buried. Month after month, the very foods he’d chosen to protect his heart were the ones straining it — until the day his condition worsened and he was admitted to hospital. That broke something open in me again. He had wanted to choose well. He had genuinely tried. The label simply gave him no honest way to tell the safe choice from the harmful one — and of everyone, it failed the person trying hardest to be careful. An unreadable label isn’t a harmless little exaggeration: for someone whose heart depends on what they eat, it can be the difference between a good choice and a hospital bed. People who want to choose healthy deserve a way to actually tell what’s healthy.

He had wanted a way to tell the safe choice from the harmful one — and so did every other family, still left in the dark: not because they didn’t care, but because no one ever gave them the key. Knowing what’s in your food and on your skin shouldn’t be a privilege for the few who can read the fine print — it should be a right. I couldn’t teach 1.4 billion people to read labels the slow, painful way I did. But I knew how to teach a machine to do it — I’d spent a decade doing exactly that for sputum, blood, eyes, symptoms and diet. There was only one difference: every one of those tools was built for a doctor or a technician. Decode Label is the first time I’ve put that same rigour directly into your hands. So I pointed everything I knew at the label. Decode Label reads any product — food, skincare, supplements — the way I once taught AI to read sputum: every ingredient checked against India’s own FSSAI rules, against EU and WHO safety data, against the carcinogen and allergen lists — and turned into one honest number from 0 to 10, in plain language, with the why behind it. And above every line of that code sits one rule, the same one that started this journey: no brand can ever pay to change its score.