‘I still have my sons learning how to code’

Gabe Monroy, VP of developer experience at Google Cloud, is spearheading a transformation in software development. His focus: Applying generative AI to the software development lifecycle through Google’s Gemini Code Assist.

“We are seeing a ton of interest in some of the unique features that Gemini Code Assist offers,” Gabe says, reflecting on recent interactions with developers and companies in India. He cites a startup called Adya, which is leveraging Google’s tool to help enterprises build applications faster on ONDC (Open Network for Digital Commerce). “The CTO of the startup described how they’re seeing about 50% productivity gains across their entire software development lifecycle,” he says.

From requirements generation and database design to UI/UX creation and backend coding, Gemini Code Assist is showing promise across the board. In some areas, such as backend coding and acceptance test writing, the productivity increases are over 50%.

Despite this, Gabe remains optimistic about the continued importance of human programmers. “I still have my sons learning how to code,” he says, “though I suspect the coding that they might do a few years from now is going to look and feel very different from the coding that I was taught when I was going to university.”

This shift in the way coding is done, Gabe believes, could lead to more applications being built. “There are more applications that need to be built in the world than there are developers and skilled people to build them,” he says.

Unique features
There are a host of coding assistants today, including the popular GitHub Copilot. Gabe says two key features set Gemini Code Assist apart. First is its large context window, which allows it to process up to 100,000 lines of code. “When you think about the kind of work enterprises do, they are really focused on large code bases,” Gabe says.

The second is a feature called code customisation. Unlike other coding assistants that rely solely on public data sets, Gemini Code Assist can securely index a company’s private code. This allows it to incorporate proprietary methods and practices into its AI suggestions, a capability that Gabe says is unmatched.

Looking to the future, Gabe sees potential in semiautonomous agents that can work towards achieving a desired intent with minimal human intervention. However, he stresses that for such solutions to be truly useful in enterprise settings, they need to integrate seamlessly with existing systems and best practices.


GN Awards 2025: Vote for your favorite Gadgets