The idea of building a company stayed with Ganesh Gopalan for decades before it became reality. As a student, the future co-founder and CEO of Bengaluru-based Gnani.ai was influenced by Dilbert Future, a book by Scott Adams that encouraged readers to think differently about careers and opportunities.
He began maintaining a diary, recording monthly and annual goals, with entrepreneurship featuring prominently among them. Yet the opportunity to act on that ambition took years to materialise.
After completing a degree in computer science from the University of Mumbai, Gopalan spent several years in consulting in the US before returning to India to pursue an MBA at the Indian School of Business as part of its founding batch.
The experience exposed him to people willing to take risks and build institutions from scratch. Later, at Texas Instruments, he became interested in creating technology products with broad societal impact.
By then, India had already produced several successful services and software companies. What interested Gopalan was something different. He believed India had yet to demonstrate that it could consistently build globally competitive deeptech products from the ground up.
That ambition eventually brought him together with Ananth Nagaraj, a colleague at Texas Instruments. The two began discussing what kind of company they wanted to build. They were clear on one point: if they left corporate careers, it would be to pursue a difficult technology problem rather than a conventional software business.
“We started when deeptech and AI were unfashionable,” Gopalan recalls. The founders became convinced that voice technology represented a large unsolved problem. Human-machine interaction was becoming increasingly important, and they believed voice would eventually emerge as a primary interface.
The challenge was complex enough to justify a deeptech approach and large enough to create a meaningful business if solved.
Intensive customer research
The decision was not made overnight. The pair spent months speaking with customers, researchers, industry leaders and AI practitioners. Many conversations happened while travelling together for work. They gathered use cases, studied market gaps and tried to understand whether a company built around voice AI could compete globally from India.
The feedback was often discouraging. Friends advised caution. Industry experts questioned whether an Indian startup could compete against better-funded international research labs. Some dismissed the idea altogether. Over time, however, the founders began separating information from opinion.
Customer conversations and market data helped refine the opportunity. The scepticism, meanwhile, became easier to ignore.
“Every conversation helped us learn more about the business and technology, but the final insights were our own,” Gopalan says. In 2017, they launched Gnani.ai. The company develops voice-first generative AI systems for enterprises, including speech recognition, conversational AI, voice agents and speech-to-speech models. Its technology is used by businesses to automate customer interactions across multiple Indian and global languages.
The early days were marked by uncertainty. The founders initially worked from the office of former Texas Instruments colleagues who had started companies of their own. They hired their first employee, who remains with the firm today, before gradually expanding the team and moving into a larger office. While the core vision remained unchanged, almost everything else evolved. Products were refined, technologies rebuilt and business models adjusted. Gopalan describes the journey as involving “a few thousand iterations”.
Raising capital proved equally challenging. Frontier AI attracted little investor interest at the time, and the company accumulated hundreds of rejections. Investors often appreciated the founders’ technical ambitions but questioned whether deeptech innovation could emerge from India.
A breakthrough came unexpectedly. After being introduced to an investor in the US, the founders were advised to speak with Samsung Ventures, which was evaluating similar technologies. The engagement led to a reassessment of their own capabilities.
“When Samsung benchmarked our patents and technologies, we realised our voice AI models compared very well globally,” Gopalan says. The validation helped attract further backing. Gnani.ai has since raised about $18 million from investors including Samsung Ventures, Info Edge Ventures and Aavishkaar Capital.
The business has also begun translating technological progress into commercial results. According to startup intelligence platform Entrackr, the company reported revenue of Rs 53.87 crore in FY25, more than doubling from Rs 23.09 crore a year earlier. It also turned profitable, reporting a net profit of Rs 3.19 crore compared with a loss of Rs 51 lakh in FY24.
Today, Gnani.ai processes more than 30 million voice interactions daily across over 12 languages and serves more than 200 enterprise customers. Among its recent milestones, Gopalan points to processing 23 billion inferencing tokens in a single day across its models. The company was also selected under the IndiaAI Mission as one of the first three firms tasked with building sovereign frontier AI models. Another milestone was the public launch of its speech-to-speech models in the presence of the Prime Minister.
Looking back, Gopalan believes the hardest part was not building the technology but making the transition from a stable corporate career to an uncertain startup journey.
If there was one mistake, he says, it was underestimating what the team could achieve. “It took time to realise we could build technology that was better than what much larger companies were building, provided we stayed focused,” he says.