At a time when global tech CEOs around the world are encouraging their engineers to adopt as much AI as possible to get basic coding done, Zoho Corporation’s Chief Scientist, Sridhar Vembu, has presented a slightly different take on how his company deals with the inclusion of AI tools at workplace. Vembu has confirmed that the Indian SaaS giant is now utilising AI coding tools across its entire company, even though it doesn’t ‘force feed or mandate’ it across the teams.
Vembu stated clearly that all AI-assisted work carries a non-negotiable condition for every engineer. “We use AI coding tools across the company, with the caution to engineers that they have to review and approve all the code and take responsibility for it. We don’t force feed or mandate AI tools, and leave the decision on how best to use AI to experienced engineers,” stated Vembu in a post on X.
Sridhar Vembu says AI leads to dramatic productivity gains
The results of Zoho’s AI adoption have been said to be tangible, especially among its most seasoned personnel. In his post on X, Vembu shared examples of efficiency, highlighting an account of a senior engineer who is responsible for performance-critical UI work. Vembu says that this veteran developer, who has decades of experience, reported a major gain in his output.
“He told me that he is now shipping features that would have taken him 3 weeks of work that he got done in a day,” says Vembu.
The technology’s impact extends to less experienced staff as well, with Vembu stating that “One of his less experienced team members has built an internal tool with AI that is being used across other teams. He said that team members could not have built the tool without AI.”
For UI development specifically, Vembu is “very bullish on AI,” recognising its effectiveness in this domain.
Vembu urges responsible AI adoption
With Vembu’s latest post, Zoho’s measured approach to AI adoption is more prominent than ever. Zoho’s culture of AI adoption contrasts sharply with the sweeping AI mandates seen at rivals like Google and Microsoft, who have reported tup to 30 per cent of their new codebase is now AI-generated. Vembu’s suggestion is to avoid the technical debt and security vulnerabilities that can accumulate from unverified code.
However, Vembu clarifies that his views are based on collective experience. “All these are independent of my own research project. Just reporting on our collective experience,” states Vembu by the end.
