He Watched Hardware Die Twice. Now He’s Building the Anthropic of Factories.

Avnish Kumar found out that the Prime Minister of India had mentioned his company on national radio because his father called him on a lazy Sunday afternoon to tell him.

Nobody at LivNSense had any advance warning. The company was still early-stage in 2023, running a handful of pilot deployments, when it took on a small government project detecting pipeline leaks in a 200-household colony in Rohtak district, Haryana. The technical problem was genuinely hard. Most underground pipeline maps in India don’t exist or can’t be trusted, so the team had to infer leaks with AI rather than working from known infrastructure. It worked the first time, which Kumar says almost never happens.

Months later, Narendra Modi mentioned the project in Mann Ki Baat. What struck Kumar more than the mention itself was what happened after it. The Prime Minister’s office followed up for weeks, tracking press coverage, pulling user feedback, coordinating a Doordarshan crew to film a proper segment. “Their only expectation,” he says, “was that the startup actually benefits from it.” The credibility never faded. Years later, pitching an oil and gas CEO in the UAE, Kumar listed off LivNSense’s industrial credentials and got a polite but unmoved response. Then he mentioned the Prime Minister. “He went completely silent,” Kumar says. “That’s when he got convinced.”

Watching the Same Collapse Twice

Kumar spent 25 years inside Indian IT and telecom, including stretches at Infosys, Wipro, HCL, and Tech Mahindra. He watched the hardware business in telecom get hollowed out by software starting around 2010, and then watched almost the same thing happen in semiconductors. Both times, people around him insisted it couldn’t happen to their part of the industry. Both times, it did.

So when he looked at manufacturing, still entirely dependent on hardware and still pricing everything around capital equipment rather than intelligence, he recognized the pattern a third time. “If you are not using intelligence,” he says, “how are you going to save energy, how are you going to save people’s lives?” That conviction, more than any specific business plan, is what pulled him out of a comfortable corporate career in his forties.

The Day He Couldn’t Eat Lunch

Kumar had never actually set foot on a factory floor. His entire career had happened in what he calls the air-conditioned rooms of the IT industry. His first real plant visit came through a partnership with a German firm, touring a heavy equipment manufacturer near Pune that built large excavators.

He watched workers welding for eight-hour shifts, breathing in fumes the whole time. “I was feeling like, how can a human survive this, and for how long,” he says. The exposure affected him so directly that his sense of taste changed for the rest of that day, and he couldn’t finish his own lunch. It became the reason LivNSense’s first product targeted welding safety and efficiency at all, and it planted something else in his head too: better automation could also remove a reason the work had always excluded women, since the job no longer had to depend on tolerating conditions like that.

Failing, Getting Rejected, and Learning Chemistry at 45

That first product, called WeldSense, failed commercially. The technology worked, but it cost more than the manual labor it was meant to replace, which in India at the time made it a non-starter. Kumar treated it as information rather than defeat, and went looking for an industry further along in its willingness to pay for intelligence. He landed on oil and gas, and started working his old professional network to find a way in.

The reception was blunt. A prospective partner told him outright that he and his team were computer engineers with no real grasp of chemical engineering, and asked how they expected to do this at all. Kumar’s answer was that the chemistry itself wasn’t the hard part. Seventy years of industrial process knowledge was already proven and documented. What hadn’t been done was encoding it into software that could apply it continuously. He got one shot, on the condition that he wouldn’t charge if it didn’t work. The first year was genuinely difficult. He had to relearn chemistry from the ground up in his mid-forties, after two and a half decades in an unrelated field.

He’s distilled that period into three rules he still runs the company by. Stay humble enough to keep changing as a person. Never stop learning, because a single day without it is a day you fall behind. And stay at the front edge of the technology itself, rather than defending whatever you already built.

The Marriage Behind the Company

Priyanka Kumar, Avnish’s wife, has been part of LivNSense since before it was a company. For the first two years, he bootstrapped it by consulting on the side while she ran daily operations and built the early product. He only committed fully once the pandemic hit, building on top of the foundation she’d already laid.

He’s unusually direct about why industrial startups take so long to pay off. Unlike consumer or fintech products, a single successful pilot doesn’t open doors the way a case study usually does. Every new customer still wants to run their own pilot from scratch, and the company is effectively competing against incumbents like ABB and Siemens with a century of accumulated engineering behind them. Both he and Priyanka have kept personally reinvesting their own savings into the business even after raising outside capital, seven years in. What has actually held the team together through that, he says, isn’t the founders alone. It’s that the people who stayed were mostly already known to them from working relationships twenty years earlier, not hires who joined cold.

What’s Changed, and What He Thinks Comes Next

Some of the oil and gas opportunity LivNSense built years ago for the US market, and couldn’t fully sell at the time, has recently come back to life in the Gulf, as regional operators shift from pure extraction economics toward genuine cost and efficiency discipline. The underlying AI has also moved fast enough that work which took two years to build five or six years ago now takes under six months.

Kumar’s bigger bet is about something less visible than any single deal. A generation of plant operators who’ve spent twenty to forty years accumulating the tacit judgment that actually runs heavy industry is heading into retirement, with maybe five years of runway left in Europe, a handful more in the US, and somewhat longer in India. He thinks almost none of that knowledge has been captured anywhere outside those people’s own heads. “Why can’t we be another Anthropic in our sector,” he says, arguing that most of today’s agentic AI hype is missing the actual physics and chemistry that real industrial decisions depend on, which is exactly the part his company has spent seven years building. His own ranked bet for where this matters most in India isn’t the sectors getting the most attention. It’s steel and mining, where he says import-dependent coal quality alone now drives 60% of production cost, not the more fashionable bets like space technology, which he thinks are years away from any real return.