Sridhar Vembu charts Zoho's AI future with hyperintelligent systems and lessons from China

In late January, when Sridhar Vembu announced his decision to step down as CEO of Zoho Corp, the news came as a surprise. At the same time, his new role as Chief Scientist left many unclear about his focus. At Zoho's annual analyst event, ZohoDay25, held this year in Austin, Texas, he shared his assumptions about the future of artificial intelligence and its implications for businesses and costs.

Calling the current phase of technological change an existential question for software engineers, Vembu said, "This is the kind of thing that requires a lot of concentrated effort to figure out—both from a technology standpoint and in terms of its business and broader implications. So, I wanted to focus on it full-time."

Zoho has also launched a new suite of AI-powered agents built on its foundational AI, Zia. The Zia Agents will be rolled out in the coming months and will include task-specific offerings such as an Account Manager Agent, HR Agent, Customer Support Agent, and IT Help Desk Agent. Additionally, the company has introduced Zia Agent Studio, a low-code/no-code platform that allows customers to build autonomous agents tailored to their specific needs. These Zia Agents can be deployed in third-party applications or within the Zoho ecosystem. As Chief Scientist, Vembu now plans to take a hands-on approach, undertaking deep R&D in AI while leveraging Zoho's engineering expertise, proprietary data centers, and shared data models to develop value-driven, customer-centric solutions.

Deep Seek and Learnings for India

Calling Deep Seek a breakthrough that challenges assumptions about big tech's AI dominance, Vembu pointed out how China has developed powerful, low-cost reasoning engines. He believes this will drive down service provider costs for deploying agentic systems.

"I have come to the conclusion that you won’t be able to charge a lot for these agentic systems. Maybe there will be a small premium, or maybe no premium at all," he said.

While the prohibitive cost of chips and GPUs may have kept many Indian companies from entering the AI race early, Vembu suggested that India could learn from China’s approach to innovation. Despite ongoing debates, China has trained Deep Seek using over 60 AI research labs, proving that major advancements don’t always require massive capital investment.

"It tells you they’re not just throwing money at the problem. That’s where India can learn—we don’t have to pour money into innovation; we can take a smarter approach," he added.

Future of Programming

Programmers have been at the forefront of AI adoption, particularly those working on user interfaces. With AI now capable of generating code, Vembu believes programmer productivity will increase dramatically as AI systems become hyperintelligent.

"My personal estimate is that 10x productivity is possible—and 100x when the systems reach hyperintelligence," he said.

As an area of personal interest, Vembu added that there is significant room to develop tools to enhance AI-generated code. He explained that his current focus at Zoho is on integrating AI with compilers and large language models (LLMs) to optimise output.

"We need to combine AI with compilers. We must marry what we see as LLMs with our traditional technologies," he noted.

Discussing AI’s impact on enterprise software pricing, Vembu suggested that the real value of enterprise software will come from privacy, trust, and compliance, with licensing models potentially evolving into insurance-style pricing.

On the coding front, he outlined three stages of AI evolution:

1. Enhancing user experience by integrating contextual business data and improving features.

2. Enabling AI agents to take actions in the real world, such as purchasing a ticket.

3. Unlocking massive productivity gains through automation.