Coding with AI: A CTO's View
I wrote my first lines of assembly code when I was just 11 years old. Over the years, I've navigated through languages like C, C++, and Cobol, eventually moving into building enterprise-scale solutions with modern frameworks in C#, PHP, Node.js, Python, Java and Go. Today, while my role leans heavily toward strategy and leadership, I'll always consider myself a coder at heart. This journey from low-level programming to high-level architecture has shaped my belief that AI isn't a disruptive force but rather a natural progression in the way we develop technology.
Shifting the Developer Mindset
Early in my career, my approach was roughly 80% coding and 20% planning. Productivity felt directly tied to how fast and how often my fingers touched the keyboard. Yet, this approach came with a significant downside; I became emotionally attached to every piece of code I wrote. Changing requirements led me to create patchwork solutions rather than refactoring or redesigning effectively.
A significant shift occurred when I realized that my most impactful contributions occurred long before any code was written. Today, my ratio has inverted: I spend roughly 80% of my time thinking, designing, and strategizing, and about 20% actually coding. By the time I start coding, I've already envisioned the solution. This shift not only improved my productivity but also freed me from attachment to specific implementations, allowing me to adapt and refine my approach without hesitation.
The Search Query Paradox
There’s an interesting parallel between coding and something as seemingly simple as searching the internet. Two people with the same access to information can yield wildly different results based solely on their ability to formulate the right questions. Similarly, two developers tackling the same problem can arrive at vastly different solutions based on subtle nuances in their initial thought processes and assumptions.
This insight directly translates into today's world of AI-driven development.
Prompt Engineering: Coding's New Frontier
AI has democratized software development, making coding accessible to many more people. But it has also made something else clear: the quality of AI's output is directly tied to the quality of our input. Prompt engineering - the skill of knowing precisely what and how to ask - is quickly becoming the key differentiator between mediocre and outstanding results.
Just as experienced developers once distinguished themselves through careful planning and structured thinking, those same analytical skills now translate directly into the ability to effectively leverage AI.
Evolution, Not Revolution
As a CTO, I view AI in the same light as advanced IDEs, libraries, or automation tools that preceded it. Whenever my team members are stuck on repetitive tasks, I introduce tools to streamline their workflow and boost productivity. AI simply represents the latest advancement in a continuous evolution of developer tools.
Recommended by LinkedIn
Historical examples reinforce this pattern. Calculators reduced the need for manual computing. Personal computers reduced the reliance on typists and manual bookkeeping. Robotic assembly lines reduced manual labor but increased the demand for skilled technicians and engineers. Each technological leap has consistently elevated human contributions to higher levels of abstraction and creativity.
AI follows this same trajectory. By automating routine coding tasks, developers can increasingly focus on design, user experience, and strategic problem-solving - areas where human creativity and judgment remain irreplaceable.
Rethinking Interviews in the AI Era
For years, I've interviewed developers by evaluating their problem-solving skills, architecture insights, and thought processes, rather than merely testing their coding abilities under pressure. With AI-assisted development, the same core competencies apply: developers now articulate problems clearly, break them down logically, and then use prompts to generate implementations.
The core skills critical thinking, problem decomposition, and clear communication remain unchanged, reinforcing the value of foundational programming and analytical skills.
Embracing Our Future
The fundamental nature of software development is still very much intact. We must understand problems deeply, design effective solutions, and implement them efficiently. We must always think before we code and communicate clearly with both machines and colleagues.
What’s evolving is our toolkit. AI tools empower developers to focus more on the creative and strategic aspects of software development. This isn't just beneficial for business efficiency; it's immensely rewarding for developers who prefer spending their valuable time solving challenging problems rather than repeating monotonous tasks.
As technology leaders, we face a choice: fear AI as a threat or embrace it as an opportunity to elevate our teams and our own potential. I choose optimism. The same skills that distinguished great developers in the past - logical thinking, strategic planning, and clear communication - are exactly those that will define success in this AI-enhanced future.
The tools evolve, but the mindset remains constant. Let's embrace this progression and use it to build better software, deliver greater value, and solve increasingly meaningful problems. Our future isn't about replacing human coders with machines; it's about amplifying human creativity with intelligent tools.
That's a future I'm excited to build and it's one worth coding for.
Well-articulated take on a rapidly shifting landscape. The CTO role today demands not just technical depth, but also foresight in navigating ethical AI, cross-functional leadership, and cultural transformation. Exciting times ahead.
Agreed! However, I’d invite your thoughts on one trend. Within my network, my peers are noticing that there’s stratification in how coders of different experience levels leverage AI. AI saves me countless hours every week. But it’s usually a starting point. In AI workgroups we’ve been noticing that younger coders are far more dependent and frequently “vibe code” ie nudge the LLM repeatedly until arriving at an outcome that compiles. In many ways the advent of LLM/IDE tooling takes me back to the explosion of stack overflow — and therefore nothing seems new. However, even with stack overflow, one must still exercise the hard skill of synthesizing thoughts into code. Rarely can one fully copy an answer. Frequently one must rewrite a bit. Still more frequently, S.O. merely gives one the starting point to search. While indeed it does seem that we’re heading towards a world in which white collar labor supervises AIs, who among us really trusts a supervisor who can’t step into the role of their subordinate? I wonder if some of the tooling is delaying or even suppressing hard skills in the next generation of coders? How might leaders sponsor and engender healthy growth among younger devs, including in emerging and alluring AI tools?