When Surojit Chatterjee first encountered artificial intelligence (AI), it wasn’t through a slick app or a polished API. It was in the early 2000s, in a lab at IIT-Kharagpur, where a classmate and he painstakingly built a neural network from scratch to recognise speech—long before Alexa or Siri were household names. “It didn’t work that well," he admits, “but we learnt a ton."
That spirit of curiosity and grit has defined Chatterjee’s journey through the world of AI. From early experiments in speech synthesis to building neural networks for autonomous navigation at India’s Defence Research and Development Organisation (DRDO), his path has been anything but conventional.
“In my third year at IIT, I interned at the Center for Artificial Intelligence and Robotics in Bengaluru, part of DRDO. I built neural networks to help robots navigate. It was very early work—what we now see in autonomous vehicles in San Francisco, I was working on early versions of that."
From there, Chatterjee’s career took him to the US, where he worked on handwriting recognition research funded by the US Postal Service, and later to Google, where he helped build the mobile advertising business from scratch—scaling it to a $50 billion revenue engine. He also served as chief product officer at Flipkart, then helped lead Google Shopping (products ad business) to $25 billion in revenue, and played a key role in Coinbase’s growth and IPO.
But it was after decades in Silicon Valley that the idea for his most ambitious venture yet began to take shape. “I realised that, even in the best companies in the world, 50 to 60 percent of people’s time was spent just keeping the lights on—mundane, repetitive work," he says. “What if we could delegate that to AI?"
That question led to the founding of EMA—Enterprise Machine Assistant—in April 2023. The vision? To build “AI employees" that can take on operational tasks across industries, freeing up human teams to focus on creativity and strategy.
What began as a solo project—Chatterjee tinkering on his laptop, reading research papers—quickly evolved. Some of the researchers he reached out to are now part of EMA’s team or advisory board. “At the time, ‘agentic AI’ was not even part of our vocabulary," Chatterjee recalls. “And our vision was to build software that works like synthetic humans—AI agents that can plan, act and collaborate like real teams." Fast forward to today, EMA boasts over 120 employees, many of them ex-Google employees and AI experts from top tech firms.
But building EMA wasn’t without its challenges. Hiring top-tier AI talent, securing early customers, and building enterprise-grade products from scratch tested the team’s resilience. “Startups are not for the faint-hearted," Chatterjee says. “It’s like a war zone."
One of EMA’s first clients, Moneyview, a digital lending platform in India, reached out to Chatterjee’s team for help in automating their customer support. Another early client was Envoy Global, a major US immigration law firm, which reached out to EMA to streamline complex legal workflows. These early wins validated the product’s potential—and its versatility.
Chatterjee is especially bullish on AI adoption in emerging markets like India. “Indian companies are newer, more tech-savvy and the workforce is younger," he notes. “Sometimes it’s actually harder to convince older, more traditional companies elsewhere. It takes time for them to understand what this technology really is."
Still, he acknowledges the fears that come with AI—about data privacy, job displacement and ethical use. Chatterjee spends a lot of his time, educating customers—everyone is talking about agentic AI now, so there is a lot of confusion among traditional companies, about what is real and what isn’t. “Often there is apprehension around the backlash of job losses for implementing AI at large scale," he says. “When computers came in, everyone thought they will lose jobs. When, in fact, they created many more jobs. I think the apprehension around AI is similar. But I do think this is the next industrial revolution."
To address these concerns, EMA has embedded responsibility and trust into its foundation. The company achieved SOC 2 compliance (evaluates a company’s systems and processes) within six months, hired one of the world’s top ethical hackers to test their systems, and built a robust governance layer before writing a single line of product code. “You can’t do this retroactively," he says. “Security and responsible AI have to be built in from day one."
Looking ahead, Chatterjee sees AI transforming every corner of the enterprise—from HR and health care to legal and logistics. “Every company will be leaner. But that doesn’t mean fewer opportunities. It means more companies, more innovation, and more people doing meaningful work," he says.
First Published: Jun 24, 2025, 13:55
Subscribe Now