Exclusive: GitHub's Sharryn Napier on Agentic AI reshaping developer productivity

In just three years, GitHub has gone from a code-hosting platform to a driving force behind a seismic shift in how software is developed.

At the heart of this change is agentic AI - a more advanced evolution of AI that positions developers not just as coders, but as orchestrators of intelligent digital assistants.

"Software is a boardroom conversation, and so is AI," explained Sharryn Napier, GitHub's Vice President for Asia Pacific and Japan. "The combination of these two things is leading us to an agentic future."

Unlike early AI tools focused on simple code completion, agentic AI is about delegation.

Developers assign tasks to AI agents - like writing test plans, fixing bugs, or creating boilerplate code - and oversee the process from a high-level perspective. It's a shift from reactive assistance to proactive orchestration.

"When we released GitHub Copilot Business back in February 2023, it was very much about code completion and prompting," Napier explained. "Agentic AI is that next evolution. It's about giving the agent tasks - 'go and do this' - and supporting the developer from brainstorm to code generation."

This leap forward could be pivotal for countries like Australia, where a growing digital economy is hindered by a persistent skills gap.

Napier pointed to a projection from the Tech Council of Australia that the country could be short more than 160,000 technology workers by 2030.

"That hasn't improved at all," she said. "We have a very small population, so the question becomes: how do we make the developer and tech community more productive? Agentic AI can supercharge that."

Napier stressed that Australia's innovation potential remains strong. "There's a lot of creativity here," she said. "We see some really great Australian software companies, but innovation can only go so far when you don't have the people. AI helps us stay competitive globally."

So what exactly does "supercharge" mean in this context? Napier described it as a multiplier effect. "It's like having a team of peers. You still review their PRs, you iterate on their work, but agentic AI gives you the ability to multiply your capability."

Although still in its early stages, GitHub is already seeing real-world applications. Tools like GitHub Workspace are allowing developers to go from ideation to implementation faster than ever.

"You're able to leverage agents throughout the whole development process - brainstorming, building code, writing test plans," Napier said. "That's going to change the game."

And the results are already measurable. Major Australian banks like Westpac, CBA and ANZ have reported productivity gains after adopting GitHub Copilot. These improvements come from internal evaluations conducted by the organisations themselves, not GitHub.

"Code completion is fabulous," said Napier. "Developers are 75% happier and 55% more productive. But when AI agents are infused across the whole GitHub platform, that's when we'll see even more impact."

What were considered impossible tasks just a year ago are now being completed with ease, she added. "One of our engineers said something they thought couldn't be done last week is now happening this week."

From writing test plans to resolving code errors, the range of tasks agentic AI can handle is growing quickly. But it won't replace developers, Napier insisted.

"They're always going to be at the centre of decision-making," she said. "They understand the strategy. They know what problem they're solving. AI just helps them execute faster and more creatively."

The vision extends far beyond Australia.

GitHub's ambitions are global, particularly in rapidly advancing markets like India. Napier described a recent visit where she witnessed children coding in Hindi.

"Normally you'd have to learn English to code," she said. "Now we're dropping the barrier of entry. Kids who previously couldn't access software development can now start coding. That is massively exciting."

India, with its large base of Global System Integrators (GSIs), has become a critical player in the AI space.

These firms, Napier explained, are already deploying agentic AI to build software faster and more securely for multinational clients.

"They have a unique impact no other community can have," she said. "The scale and creativity are phenomenal."

As agentic AI embeds itself deeper into workflows, Napier sees the developer role evolving significantly.

"We're moving from pair programming to orchestration," she said. "Developers become the directors – the conductors of an orchestra of agents."

This shift could also redefine job satisfaction. According to GitHub's internal data, developers using AI tools are not only more productive but also markedly happier. "Seventy-five percent happier," Napier said. "Coding is fun again."

As software development grows more complex, AI could alleviate some of the cognitive burden on engineers, helping them focus on creative problem-solving rather than repetitive tasks.

"We've seen that when developers are happier, they're more likely to stay in their roles," she said. "It's not just about productivity. It's about satisfaction."

For businesses unsure how to prepare for agentic AI, Napier's advice was simple: "start small, and prioritise literacy."

"Jump in. Don't be scared to try things," she urged. "We need to teach people how to prompt, how to interact with AI, and how to leverage it. Just like data literacy, AI literacy is essential."

She cited Indian GSIs that have mandated AI education across their organisations. "That's what other companies need to do. Because this isn't coming – it's already here."

As GitHub pushes forward with its audacious goal of reaching a billion developers worldwide, Napier believes agentic AI is the key.

"We're at a moment in time where even people who don't know how to code can start building software."