Why I Rely on AI as the Backbone of My Development Workflow
After more than two decades building web applications, my approach to software development has evolved dramatically. Today, AI isn't just another tool in my belt—it's the foundation of my workflow. I want to share openly how I leverage AI, why I take this approach, and why I believe it leads to better products for users and greater satisfaction for developers.
Embracing the Modern Toolkit
When I started coding, "developer productivity" meant becoming faster at hand-writing and refactoring large codebases. Patterns and boilerplate ruled my days. Fast forward to now: intelligent automation lets me offload that repetitive work to machines. Tools like GitHub Copilot, Perplexity, Cursor, and modern conversational voice prompting have fundamentally changed how productive—and creative—I can be. Instead of relying solely on text-based queries, I can now express requirements to AI in natural language, brainstorm solutions out loud, and iterate on code in real time. I also leverage screenshots to provide visual context, web links to reference documentation or examples, and even other AI-generated prompts to build richer, more precise instructions. These tools and techniques allow me to architect, debug, and refine software with greater speed and precision, giving me more runway for deep thinking and impactful improvements.
I no longer view my role as a developer as manually producing every line of code. Rather, I architect the requirements, translate them into effective prompts, and review/validate the outputs generated by AI. I use my experience to identify weaknesses, test boundaries, and make targeted manual edits when needed. This lets me focus on what really matters: design, correctness, performance, and user experience.
The Workflow: AI-First, Human-Owned
Here's how my process usually plays out:
Architecting & Prompting: I break down requirements, ask the right questions, and use AI to brainstorm solutions and scaffolds.
Generation: AI creates the first pass of the code—models, endpoints, components, tests, etc.
Review & Enhance: With a critical eye, I review and iteratively improve what AI produces, leveraging my past experiences with similar patterns.
Manual Testing: I go beyond "does it run?" and apply domain knowledge to challenging use cases, edge conditions, and performance bottlenecks.
Peer Review: The code is subjected to review by skilled colleagues—not just checking for bugs, but also maintainability, clarity, and purpose.
The result isn't "AI-written code"; it's a product shaped and delivered by a deeply accountable human who uses AI as a force multiplier.
Why Not Just Hand-Code It?
One question I sometimes hear: "Can't you just hand-code this segment to show your raw thinking?" My honest answer—after years of doing just that—is that it's no longer a good use of time or talent. If the goal is originality or understanding, I can walk through the code, explain every choice, and refactor on the fly. But recreating old patterns line-by-line, when AI can accelerate the journey, feels like deliberate inefficiency.
I believe our industry should reward judgment, not keystrokes. My focus is on delivering value—faster and at higher quality—by harnessing the best tools available.
My Philosophy in a Nutshell
AI is not a replacement for ownership; it's a catalyst for better ownership.
Pattern recognition and repetitive coding are best handled by machines.
My expertise is in prompts, architecture, design, code review, and delivering quality—not in avoiding automation for its own sake.
I'm always happy to explain, validate, and iterate on my solutions—just not interested in reverting to manual workflows to prove skill.
Let's Build Smarter
Development is changing rapidly, and I'm embracing that change head-on. If you're curious about the results an experienced, AI-first developer can deliver—or want to talk about building smart, modern software—I'd love to connect.