The real battle in artificial intelligence is shifting from building bigger models to ensuring they behave predictably, as enterprises prioritise reliability while adopting AI, said Satya Nitta, chief executive and cofounder of Emergence AI, a startup focused on building agentic AI systems.
“Today’s AI agents are probabilistic and loosely governed. That is why enterprises are cautious,” Nitta said, adding that the missing piece is determinism. “If you want these systems to operate in real-world environments, especially critical ones, you need predictable behaviour and guarantees,” he said.
Against this backdrop, the company is expanding its research footprint in India with the launch of Emergence India Labs in Bengaluru. The facility will focus on building autonomous agents rather than large language models and is expected to scale to around 500 researchers and engineers over time.
While the race to build foundation models is concentrating among a few large players such as OpenAI, Google and Anthropic, the next phase of competition will be around the control layer that sits on top of these models, said Nitta, who was the global head of the cognitive sciences department at IBM Research before founding Emergence AI with two others.
“The real shift is towards building systems that can take decisions and act reliably, not just generate responses,” he said.
Emergence AI builds software that uses multiple AI agents working together to complete tasks across enterprise systems, including analysing data, making decisions and triggering actions. Its platform is designed to coordinate these agents, manage workflows and integrate with existing enterprise tools.
The New York-headquartered AI research and product startup is currently at a Series C stage and has raised about $97.2 million across two funding rounds, with key investors including Learn Capital.
The company said its Bengaluru facility is meant to accelerate the nation’s shift from IT services to autonomous systems and advanced manufacturing.
Siddhartha Gadgil, professor at the Indian Institute of Science (IISc), will join as chief scientist while maintaining his academic affiliation.
Unlike multinational R&D outposts that operate as satellite extensions of overseas global headquarters, Emergence India Labs is conceived as a core, AI-native R&D epicentre – designed to anchor sovereign AI capability, the company said.
“This is not a satellite centre. The idea is to make India a decision-making hub for frontier AI work,” Nitta explained. “A lot of the talent exists here, but the innovation often happens outside. We want to anchor that in India.”
The lab will work on deploying agent-based systems in sectors such as telecom, financial services and digital infrastructure, where systems are required to operate with high levels of consistency. “These are environments where systems have to behave predictably and consistently,” Nitta said.
“Today’s AI agents are probabilistic and loosely governed. That is why enterprises are cautious,” Nitta said, adding that the missing piece is determinism. “If you want these systems to operate in real-world environments, especially critical ones, you need predictable behaviour and guarantees,” he said.
Against this backdrop, the company is expanding its research footprint in India with the launch of Emergence India Labs in Bengaluru. The facility will focus on building autonomous agents rather than large language models and is expected to scale to around 500 researchers and engineers over time.
While the race to build foundation models is concentrating among a few large players such as OpenAI, Google and Anthropic, the next phase of competition will be around the control layer that sits on top of these models, said Nitta, who was the global head of the cognitive sciences department at IBM Research before founding Emergence AI with two others.
“The real shift is towards building systems that can take decisions and act reliably, not just generate responses,” he said.
Emergence AI builds software that uses multiple AI agents working together to complete tasks across enterprise systems, including analysing data, making decisions and triggering actions. Its platform is designed to coordinate these agents, manage workflows and integrate with existing enterprise tools.
The New York-headquartered AI research and product startup is currently at a Series C stage and has raised about $97.2 million across two funding rounds, with key investors including Learn Capital.
The company said its Bengaluru facility is meant to accelerate the nation’s shift from IT services to autonomous systems and advanced manufacturing.
Siddhartha Gadgil, professor at the Indian Institute of Science (IISc), will join as chief scientist while maintaining his academic affiliation.
Unlike multinational R&D outposts that operate as satellite extensions of overseas global headquarters, Emergence India Labs is conceived as a core, AI-native R&D epicentre – designed to anchor sovereign AI capability, the company said.
“This is not a satellite centre. The idea is to make India a decision-making hub for frontier AI work,” Nitta explained. “A lot of the talent exists here, but the innovation often happens outside. We want to anchor that in India.”
The lab will work on deploying agent-based systems in sectors such as telecom, financial services and digital infrastructure, where systems are required to operate with high levels of consistency. “These are environments where systems have to behave predictably and consistently,” Nitta said.
