Twenty years ago, building a news portal meant hiring an army. Rediff and Indiatimes ran hundreds of content creators, editors and managers. in.com did it with around forty people in total, dev, design, content and infra, because we were a smart aggregator, and that was ahead of its time. Then I built a bigger news product over a weekend, on my own, and it reached two million people. The difference is what one person with the right agents can now do.
It started with my daughter.
Her school trip was stuck in Madrid, booked home on Etihad through the Gulf, and by the second day of the conflict that route was the whole worry. I wanted one screen with everything on it: live television, a flight radar, a map of the strikes, the emergency numbers that actually work. I went looking for it. It did not exist. The news was scattered across twenty tabs, each one a single angle, none of them talking to each other. So I built it, against the clock. I had twenty-four hours to get it live, and I made it.
I put it online and it moved on its own. Two hundred thousand people in forty-eight hours, passed hand to hand across WhatsApp and Telegram and a few IIT message boards I will never see. That told me something simple: everyone was as lost as I was. So I made it free and permanent. No ads, no sign-ups, no paywall. war.direct has now been seen by more than two million people.
One screen for a crisis
war.direct is a single screen for a conflict. Twenty-five and more live channels, side by side, from every side: Al Jazeera, BBC, CNN, Reuters, India Today, NDTV, DW, Euronews. Not one perspective. All of them, next to each other, so you can form your own view. A live strike map with more than a hundred verified attacks placed at their real locations. A flight radar tracking military and civilian aircraft across the region. Naval movement. An AI-built timeline of the war. And verified emergency numbers for hospitals, airlines and embassies across eleven countries, including every Indian mission in the affected zone.
The team is the agents
There is no team. I built it with Claude as my co-builder, and it runs on agents I built alongside it.
One agent crawls the news around the clock, from every side of the story, Iran and Israel, Russia, China and the United States, so no single narrative gets to be the only one. One reads the room, scoring sentiment as it shifts. One maps the strikes as they are reported and verified. One pulls in public video feeds from across the conflict. They run without me. I check in. I do not operate it by hand.
Twenty years ago, that was a building full of people. The big portals ran hundreds; even a lean shop like in.com needed a real team to do far less than this. war.direct does more, from more sources, faster, and the operating team is one person and a set of agents that never sleep.
in.com had the idea
At in.com we were chasing the same thing, twenty years too early. The plan was a newsroom that largely ran itself. One massive crawler sweeping the open web, using regex to catch everything that was trending the moment it trended. Sentiment analysis reading each story from every side, so you got the whole shape of it instead of one outlet’s angle. Collaborative filtering surfacing the music, the videos, the games and the rest around whatever you were reading, the way the big recommendation engines do now. The machine gathered and sorted. A person sat at the end as the final curator, the human in the loop deciding what was true and what mattered.
It worked, held together by forty people doing it largely by hand. What we never had was a way to keep that machine running without a room full of people feeding it. war.direct is the same idea, with that part finally solved.
OSINT grew up
None of the raw material is secret. The strike locations, the flight paths, the broadcasts, the chatter on the ground, all of it is open. What changed is that the work of pulling it together, the part that used to need a Palantir-sized engineering org, can now be assembled by one operator over a weekend. Open-source intelligence, plus public feeds, plus a few good agents, gets you a real-time, data-backed picture of a live war. The barrier was the labour, and the labour just collapsed.
The state noticed, briefly. The first version, war.org.in, was suspended by the registry within a day. We were back in an hour on war.direct. When the cost of rebuilding is an hour, you do not really go down.
Why a weekend matters
I built war.direct because my daughter was in the air and I could not find a straight answer, and it turned out a couple of million other people could not either. Crisis information should be free, and it should sit in one place, from every angle. That part is simple.
The other part is the one I keep coming back to, and it is the whole of what Antfarm is now. A real thing, built and run, by one operator with agents doing the work that used to take a hundred people. If a weekend project can do this in the middle of a war, the question for every company is what you would build if execution were almost free.
The barrier was the labour, and the labour just collapsed.
So build it. Then run it. If you know someone in the region, send them war.direct. And if there is something you have been waiting for a team to build, you do not need the team. Come and find us.