Coding with AI: A Take on the Future of Software Development

In this episode of the Snowpal Software Development, Finance, and AI podcast, Krish Palaniappan speaks with Michael Rispoli , CTO of Cause of a Kind , about the dramatic shifts AI is bringing to modern software development. As a hands-on CTO, Mike offers a candid, real-world view of how tools like Cursor, Zed, and Claude are transforming engineering workflows.

From Skepticism to Embrace: AI’s Evolution in Development

Mike recounts his initial hesitation around AI, drawing a parallel with the Web3 hype cycle. As a digital agency founder who had watched the crypto bubble rise and collapse, he was wary of another “next big thing.” But the last few months have changed his perspective.

The turning point? Tools like Claude-powered agentic editors have matured to a point where Mike says he hasn’t written traditional code in weeks. “It’s changed everything,” he says, explaining how the speed and accuracy of AI assistance have supercharged productivity.

Cursor vs. Zed: A Developer’s Experience

Mike dives into the real-world trade-offs between Cursor (built on VS Code) and Zed (Rust-based and lightning-fast). Both tools promise AI-assisted coding, but each has its nuances. While Cursor aims to learn your whole codebase and edit intelligently, Mike found Zed to be faster and more stable—especially in legacy projects.\

The promise of AI editors is eliminating the copy-paste back-and-forth between developer and assistant. These tools instead allow engineers to work in a single window, reviewing diffs and directing the AI with context-rich prompts.

When AI Struggles: Prompting vs. Problem Solving

But the journey isn’t seamless. Mike details the friction of relying on AI to understand complex systems. In one example, he prompted an AI agent repeatedly to make changes in a rules engine but found diminishing returns and growing frustration. “It felt like gambling,” he joked. “One more prompt and maybe I’ll get it right.”

He raises a profound question many developers now face: Should I think through the problem myself, or let the AI think for me?

Legacy Code Is Back in Vogue

Contrary to common belief, Mike argues that AI may be more effective in mature, legacy codebases than Greenfield projects. Why? Older apps offer rich context, consistent patterns, and better test coverage—much like a junior engineer learning a system. With Greenfield apps, AI tends to default to repetitive tech stacks, limiting innovation.

Non-Developers Can Build—But Should They Ship?

The AI era empowers non-developers to build MVPs and clickable prototypes. But Mike cautions that deeper technical knowledge is still essential. Without it, users often make flawed architectural decisions—like mixing blob and database storage inappropriately, or misusing third-party services.

While the prototyping barrier is lower than ever, production readiness still demands experienced hands.

Productivity Gains: 3x, 5x, Even 10x

For engineers, the gains are substantial. Mike estimates he can now complete three to ten times more work, depending on the task. Annoying and repetitive tasks—like writing config files, integrating APIs, or troubleshooting minor issues—are handled in minutes with the right AI tools.

The result? Developers aren’t blocked. Mike can juggle multiple projects, delegate common issues to AI agents, and receive completed tasks with staging links—without touching a line of code himself.

The Solo Founder Era Is Real

Mike confirms a provocative idea: one person could now realistically build a unicorn. With the right skills, a single engineer can wear multiple hats—frontend, backend, DevOps, even design—with AI filling in the gaps.

“AI makes you a specialist,” he says, allowing even generalists to produce production-grade output in areas where they’re less experienced.

Coding Still Matters

Despite the AI boom, coding isn’t obsolete. Mike draws on his own journey—from front-end engineer to full-stack CTO—to emphasize that foundational knowledge is still essential. Understanding code helps you get better results from AI, make informed architectural decisions, and avoid critical mistakes.

“Just like math teaches you how to think,” he says, “so does code.”

Conclusion: Software Development Is Becoming Human-Centric Again

AI is transforming development—but not by replacing developers. Instead, it’s elevating them, freeing time for customer conversations, product strategy, and creative problem solving. Developers can now focus less on syntax and more on outcomes, less on code and more on users.

The future, as Mike paints it, is one where a single, product-focused engineer could launch a world-changing app—with AI as their co-pilot.