On childhood, AI, and why I'm here.

Why childhood and AI matters now.

For the first time in human history, the question of what a child learns to be is being shaped, in real time, by software. Not eventually. Not in some future scenario. Right now — on phones in living rooms across India, on the YouTube tab a five-year-old has learned to open by herself, on the AI chat that another five-year-old has started asking questions to before he can fully read.

This is happening whether or not anyone is taking it seriously. Most companies aren't. The few that are tend to come at it from two unhelpful directions: as an engagement opportunity, or as a moral panic.

I think there is a third way. I think children deserve software that is actually thinking about them. That listens. That is patient. That respects how slowly a person's relationship with a language, with curiosity, with confidence, actually develops. I have spent the last year increasingly convinced that someone has to build this carefully — and that the small number of people building it carefully should not all be doing it from Silicon Valley, in English-first contexts, for children whose first language was the one their AI was already trained on.

What I observed at scale.

I spent most of my career at the consumer scale of India — at Airtel, OYO, Network18 — building and running businesses that touched tens of millions of people. The thing you learn at that scale is what I would call the operating gap: the distance between what a company says it is doing and what is actually happening to the person on the other end.

At consumer scale, the gap is usually wide. The product team talks about empowerment; the user is being upsold. The marketing team talks about trust; the funnel is engineered for retention. The customer experience team talks about delight; the call is about a billing dispute.

I do not think this is anybody's fault. The incentives create the gap. But after about a decade and a half of working inside it, I started noticing that the gap was widest at the places where the business touched the most vulnerable users. The first-time internet users. The people for whom English was a third language. And — when I started watching closely — the children.

I would sit in households where the parent's English was a second language, and watch a child try to read aloud from an app that did not understand her. And I would watch the parent quietly conclude that the child was the problem. The software was almost never the problem in their eyes. The child always was.

Why AI changes the leverage.

The reason I am building this company now — and starting it small — is that AI has quietly compressed the distance between an idea and a real thing in a real child's hands. Work that used to require a team of twenty can, in the right hands, now be done by two or three. Sometimes by one.

I do not think this is a story about replacing people. The companies that defined my career were thousand-person operations because they had to be: the supply chains, the on-ground operations, the legal complexity, the customer service load. AI does not change those. But for the work that lives upstream of all that — the engineering plumbing, the design iteration, the analytics interpretation, the everyday operational glue between ideas — the distance between what a careful person can imagine and what they can put in front of a real child by Tuesday has collapsed.

For a company whose entire premise is being unusually careful about the experience a child has with our software, that compression matters. It means the person watching the children is the same person building for them. The strategy does not get reinterpreted through three layers of team on the way to the screen.

This will not stay this way forever. We will hire as the work demands it. But while the first products are still finding their shape, the right size of company is small enough that the same person who sat with a child on a Tuesday is the one shipping the next prototype on a Friday.

Why India is the right place to build from.

I have lived almost all of my life in India and worked in industries that exist for India. I would be doing a worse job at this company if I tried to build it from anywhere else.

The children we are building for are not the children most consumer AI is built for. They grow up between two or three languages. They learn English in a school where their teacher's English may itself be a second language. They watch YouTube in Hindi in the morning, study English textbooks in the afternoon, and speak Marathi or Tamil or Bengali to their grandparents in the evening. By the time they meet an AI, they have already been doing the cognitive work of switching contexts for half their life.

That child does not need adapted-from-American-English software. She needs software built for her specifically — for her phonology, her vocabulary, the particular small confusions she will have with sounds that her family language does not contain, the patience she will need to read a book in a language her parents are also still learning.

There are millions of children like her — across India, across the Indian diaspora, and across most of the rest of the world that learns English as a second or third language. They are, in fact, most of the world's children. They are an afterthought to the AI industry. I think they should be the thesis of an AI company. So I am trying to build one.

What ANA is trying to build.

ANA PlayLabs is, in the smallest possible description, a company building software for children and the parents raising them. ZigZu is the first product: an AI reading coach that helps Indian children learn to read English the way a patient teacher would. ParentMind is the second: a tool that helps parents make sense of the chaos of advice, articles, and reels they collect about raising a child.

But the company is longer than either of these products. The honest, longer-term description is that ANA is trying to build something closer to an intelligence layer for the first ten years of a child's life — a small, careful collection of child-respecting tools that help children learn and help parents parent, without trying to replace either of them.

This is a fifteen-year project, not a five-year one. There is no rush. The children we are building for are not going anywhere. We will keep watching, keep writing what we observe, keep shipping when a child reaches for something twice — and slowly, over years, build the kind of company we would feel honoured to have built for our own families.

How I want this company to behave.

There is a particular kind of company I am trying not to build. The kind that talks about children as users, parents as conversion targets, and play as engagement. The kind that scales by stripping away the very things that made the early product worth using. The kind whose investor deck describes children as a market.

I would rather build a small, durable, slow-moving company that fewer people use, but that the people who do use it actually trust. If a parent quietly recommends us to another parent, that is the kind of thing we are paying attention to. We would rather publish what we believe and be held to it than write a privacy policy that nobody reads.

If this resonates with you — as a parent, a journalist, a possible future employee, an investor who measures companies in decades — I would love for you to follow along. The writing arrives slowly. The products take their time. The principles are written down and meant to be held against us.

I'll be here.