From Writing Code to Solving Problems: How AI is Amplifying Human Capability

Sonali Kochar has been in software development long enough to remember coding in VI editors. Over the past 25 years, she has witnessed the industry’s transformation through virtualization, Agile, microservices, and cloud computing. But AI, she says, is fundamentally different.

As Senior Vice President of Delivery and Digital Experience Practice at Encora, Kochar leads teams through this transformation with a clear perspective: it's not just about speed. It's about rethinking how developers approach their work and what makes them valuable.

You've lived through Agile, cloud, microservices, and every major shift of the past 25 years. What makes the AI revolution different from everything that came before?

The pace is unlike anything I've seen. With Agile or cloud, you had time to learn and build expertise gradually. With AI coding tools, every few months, a new capability emerges that changes what's possible.

It's not just changing what we build or how fast. It's changing who does the building and how they think about problems. That's a different category of transformation altogether.

Everyone talks about productivity gains. Give us a real example—what does this look like on the ground?

Recently, we used AI to upgrade a client's Java version from 8 to 22. What would have taken two to three months took a week. These are repeatable gains.

When new versions of Angular or Magento come in, upgrading the entire codebase used to take months. With the right prompts, you can generate 70 to 80% of the code, including unit test coverage. Something an engineer would take five to six days to write can be done within an hour.

I'm not saying all the code is perfect. There's hallucination, and you need human review. But the impact on productivity is real.

So what's the catch? What's harder now?

You need guardrails. While using AI for execution, we must ensure the code is reviewed and tested properly before it reaches production. We need to invest in governance as much as execution. You can't just trust everything AI produces.

There's a narrative that "QA is dead." What's the reality you're seeing?

It's shifted from execution to governance. We use AI to generate test cases, but we still validate them. Are they correct? Is this the right test case for that scenario?

Instead of being pure hands-on QA, we're now AI-assisted QA. AI is a tool that improves productivity, but we still need diligence: trust the code, review the code, and put governance in place.

Three years ago, what made someone a strong developer? Has that definition changed?

The fundamental skills haven't changed. We still look for comprehension, understanding of the client's needs, and analytical skills to break problems down into manageable pieces. 

You still need comprehension to describe the right problem to AI. You still need analytical capabilities to review the code AI produces. Does it solve the problem? Is this the most efficient way? 

What's becoming more critical is architectural thinking and judgment. How do systems work together? What's the human impact of what we're building? People use all these systems. AI won't have that empathy. That's where the future generation needs to focus. 

Are your developers worried about AI taking their jobs?

I haven't seen it yet. Right now, I see more excitement. People want to learn the tool and see how it can best be used. 

As leaders, we provide clarity and direction. When I started, we used VI editors. Then, IDEs generated the entire project scaffolding in 30 seconds. We became better developers. AI is another tool, the most powerful we've had so far. 

The idea is not to fear it, but to master it. We're AI-assisted developers now. AI is another collaborator and another team member. Instead of doing one thing in four weeks, we give clients ten things in four weeks. It's about leveraging it to make ourselves stronger. 

What's your advice to other leaders trying to navigate this shift?

Walk the talk. Before we give clarity and confidence to our teams, experiment ourselves. Get firsthand experience. 

Show them AI isn't replacing them. What they should fear is getting stuck in old technology, becoming outdated. That's the bigger fear. 

Encourage curiosity. Make learning a core part of delivery. Recognize and celebrate engineers who are adopting AI and sharing their approach. Let them show others: "I used AI to automate this, and this is how I did it." 

Cut through all the noise. What's the one truth about this transformation that leaders need to understand?

This revolution is not about replacing humans. It's about amplifying human capability. AI takes over execution, shifting our focus from writing code to solving problems—from execution to imagination.

The winners will be those who learn to collaborate with this intelligence, not compete or fear it.

Sonali leaves us with an analogy that has guided her through 25 years of technological change: 

Sonali's emphasis on governance, architectural thinking, and amplifying human capability offers a leadership perspective on this shift. To see how these principles translate to the technical challenges of ensuring code quality in the AI era, explore Partha Mishra's insights in Why Your Job isn't at Risk When You Own the Value Creation. Or return to the full exploration: Unlike Anything Before: How AI Has Transformed Software Development.