Sharryn Napier, VP APAC at GitHub: “AI is changing not only how people work, but who can create value with technology”

We have now had 23 “Conversations on AI“, and each one gives a fascinating insight into a different aspect of our curious industry. Perhaps none more so than with Sharryn Napier, who is the Vice President for APAC at GitHub – an organisation that embraced AI before ChatGPT shook the world.

“GitHub Copilot [its AI tool introduced in June 2022] is no longer just for developers,” Sharryn told us in this expansive interview. “Experiences such as Copilot in the CLI and the GitHub Copilot app enable teams across legal, sales, and operations to automate routine work, surface insights, and move from idea to outcome faster.”

This also means a shift in what developers are doing day to day. “[They’re] shifting their focus to architecture, design decisions, better understanding of the business problems and objectives, and the higher-judgement calls that determine whether software actually works for the people using it,” said Sharryn.

She also raises another excellent point: the explosion of code creation – “more code [is] being produced now than at any point in history” said Sharryn – means more security reviews, more maintenance, more patching.

“For boards and executive teams, that means setting explicit guardrails: approved models and data, defined decision rights, clear ownership, transparent monitoring, and controls proportionate to the risk,” she said, particularly in regard to experimentation. “Higher risk applications require stronger testing, human oversight, and evidence that outputs are reliable, explainable, and aligned with policy.”

Before we dive into the full interview, a few words on Sharryn Napier herself. As VP for the Asia-Pacific region – Sharryn is based in Sydney, Australia – she is in charge of the company’s strategy and growth. Including giving organisations insights into how GitHub’s platform can help drive their strategy and growth.

With a background in numerous global software companies, Sharryn has a wide range of experience and has helped guide her previous employers to the kind of growth she now wants to deliver to other organisations via GitHub.

All of which means that she has plenty to say about the power of AI…

Many SaaS vendors now describe themselves as “AI-powered”. What actually separates companies creating real customer value from those simply adding AI features? 

There is a big difference between bolting AI onto your existing workflows, products, and services and creating real, transformational value for customers that delivers a business outcome for them.   

If I were to borrow an example from software development, the value is less in getting AI to write your code for you and more in how it supports the entire software development lifecycle. How is it supporting planning, collaboration, code review, quality assurance, and security scanning? And is it doing this in a way that is governed and compliant?  

There are downstream consequences to focusing on one part of the process. For example, if you were to use AI to only speed up code generation, then you have created an entirely new problem on the other end for those who need to review said code. Instead of solving a bottleneck – code generation – you’ve just moved it to the other end of the process.   

This is important to note because organisations are producing more code than ever, and review, security, and governance must scale along with it. If it doesn’t, then the system defeats the purpose of deploying AI in the first place. The companies creating real value understand this because they are solving for the whole system of delivery, not one corner of it. 

If you were launching your company today with AI available from day one, what would you build differently? 

This question has a unique lens for GitHub because we were the pioneer of AI developer tools, having first launched GitHub Copilot in technical preview in June 2021. That was before ChatGPT had released to the market. So for us, it’s less about what we would do differently today, but what we would continue to do having learned so much over the years about who we are as a company and what we stand for. 

The answer comes back to our purpose as an organisation, which is to be the place where the world builds software. And at the core, it’s always been about giving software developers a choice in how they build that software. That hasn’t changed, just evolved over time. Because while GitHub pioneered AI in software development, it’s always been the home of the global developer community we serve, now 180+ million strong.  

That philosophy of choice extends to our offering today. Despite having an AI coding agent of our own, we allow developers to freely choose from a variety of models and agents that suit their needs, directly within GitHub. Where we add value is by providing the critical governance and security controls that enable developers and customer organisations to achieve outcomes securely and fast. That’s why we launched AgentHQ, an orchestration layer that transforms GitHub into an open ecosystem that unites every agent on a single platform. 

How is AI changing the role of your customers? Are you replacing repetitive work, augmenting decision-making, or fundamentally changing how people do their jobs? 

AI is changing not only how people work, but who can create value with technology. We are seeing meaningful examples of all three of these elements across development, compounding into real value for the businesses that get it right. 

GitHub Copilot is no longer just for developers: experiences such as Copilot in the CLI and the GitHub Copilot app enable teams across legal, sales, and operations to automate routine work, surface insights, and move from idea to outcome faster. That democratises capability across the enterprise, helping more people produce higher quality work, make better decisions, and contribute to innovation without needing to write code.  

The quickest, most visible change is the mundane and repetitive work moving to agents, saving hours that is redirected toward higher value work. What we’re seeing with enterprise customers is their developers are shifting their focus to architecture, design decisions, better understanding of the business problems and objectives, and the higher-judgement calls that determine whether software actually works for the people using it.  

When a developer has more context and less time consumed by repetitive tasks, their capacity frees up and the quality of their decisions improve. They can get to strategising, directing, reviewing, and validating quicker, rather than grinding the discrete code writing steps.  

This results in a compounding sequence of benefits for enterprise organisations: less manual effort spent on mundane tasks frees up resources to focus on problem solving, ideation, and innovation, which leads to better outcomes for the business.  

How do you balance innovation with responsible AI? Where do you draw the line between moving quickly and ensuring customers can trust the outputs?’ 

I do not see innovation and responsible AI as strictly competing priorities. Trust is what allows innovation to scale. The framing of speed versus responsibility was useful in the past, but less so now, because speed is a given. Every model has it, every developer is used to it, so creating code quickly is no longer the differentiator. The new success metrics are security and quality. 

There is more code being produced now than at any point in history, and all of it needs to be reviewed, secured, maintained, and understood by someone who didn’t write it. Developers are paying much closer attention to these metrics because they determine whether the speed was worth anything; volume without quality is just a larger surface area for errors. 

So the line is clear: we move quickly in experimentation, but we do not compromise on security, privacy, human accountability, or the ability to validate outputs before they affect customers or critical decisions. For boards and executive teams, that means setting explicit guardrails: approved models and data, defined decision rights, clear ownership, transparent monitoring, and controls proportionate to the risk. Low risk experimentation can advance rapidly; higher risk applications require stronger testing, human oversight, and evidence that outputs are reliable, explainable, and aligned with policy. 

Ensuring developers can solve for this duality of the AI era has been a central focus for us at GitHub. Through AgentHQ, we provide a robust harness to govern, secure, and control AI use within an organisation’s own policies and expectations. The organisation defines what good looks like, which models are permitted, what data can be touched, and what can and cannot ship. Then, AgentHQ enforces this policy, ensuring organisations can build quickly without compromising on trust, security, and responsible use of AI. 

Ultimately, speed without trust creates risk, while trust without progress creates irrelevance. The leadership obligation is to deliver both: innovate with pace, govern with discipline, and ultimately ensure people remain accountable for the outcomes. 

Beyond productivity gains, what business outcome are your customers most excited about when they invest in AI? 

As an industry, I think we can now treat productivity gains as a given. It was the dominant theme in the early years of AI, but the conversation has shifted heavily to realising business outcomes and that’s a healthy place for us to be. Customers want ROI from their AI investment and they want to be able to measure and prove it. 

ROI has historically been measured by AI adoption and time saved: how many licences are used and how many code suggestions were accepted. That told you people were using the tool, but not whether it was delivering a business outcome, which is ultimately what enterprise customers are looking to achieve. Today, they’re measuring success based on things like product innovation and delivery. Quality, security, governance and time-to-value are all nice to have, but if these benefits sit in siloes, they aren’t achieving that intended outcome. So, we look at the whole picture and focus on providing a single platform that brings them all together to make this process seamless.  

When the whole development lifecycle exists on a single platform, it is much easier for customers to measure this impact too, because all the context and evidence sits in one system. Over recent months, we have strengthened and expanded the GitHub Copilot reporting and code review and quality capabilities. For customers, the value is increasingly evident beyond individual productivity: shorter review cycles, higher-quality releases, stronger security, and faster delivery of business-critical initiatives. That value is then backed with trust with enterprise-grade governance that gives leaders greater confidence that AI adoption remains aligned with organisational policies, risk appetite, and operating requirements. 

Ultimately, tangible return on AI investment will be defined differently by each organisation. The most meaningful measures are not the volume of code produced or the number of deployments, but whether AI advances the outcomes that matter to the business: accelerating entry into new markets, bringing new products and services to customers sooner, increasing capacity for innovation, improving resilience, and reducing risk. That is the conversation we have with customers — connecting technology investment to the strategic priorities set by the C-suite and the board, and demonstrating progress through measurable business results. 

If you could give one piece of advice to another SaaS executive planning their AI strategy today, what would it be? 

My advice is to start with the business outcome, not the technology. AI capabilities will continue to change rapidly; your strategic priorities should not. Be clear about the customer problem you are solving, the value you expect to create, and the risk you are prepared to carry.  

Skate to where the puck is going. What I mean by that is this industry, and AI as a whole, is changing very fast. Not year by year, but day to day, moment by moment. If you build your AI strategy around today’s needs and capabilities, there is always a risk it could become outdated before it’s implemented.   

I have watched organisations run rigorous, multi-month strategy processes and arrive at a considered answer to a question that had already moved on. Some opportunities are short-lived and trends are fleeting. AI capabilities will continue to change rapidly; your strategic priorities should not.  

Instead, organisations should anchor their AI strategy to things that don’t change every quarter, like core business goals, mission statements, and the problems they’re solving for customers.  Build your strategy around this and iterate as things evolve.  

For boards and founders, the priority is to create the conditions for speed with discipline: invest in trusted data, clear accountability, strong governance, and a platform that can evolve as models and use cases change. Then fund a focused portfolio of initiatives, measure them against revenue, customer experience, cost, resilience and risk, and scale only where the evidence supports it. 

The strongest AI strategies are not fixed plans. They are clear strategic choices supported by an operating model that learns quickly, stops what is not delivering, and redirects capital to what is. In this environment, agility is not the absence of strategy; it is a core governance capability.

More great interviews

• Faraz Ali, Product Manager at Vertiv: “Talk to your facilities team before you finalise your AI roadmap”
• John Wright, CEO of TrueDialog: “I expect the next wave to be messaging that is accessible to AI agents”
• Adam Arellano, Field CTO at Harness: “The biggest misconception that enterprise customers have about adopting AI within business-critical software is that it is going to behave as they expect it to”
• Drew Garner, Chief AI Officer at Smartsheet: “AI will be most transformative when it reduces friction between people, systems and data”