One of the greatest privileges of my shift from corporate executive to Solopreneur is the freedom and the necessity to learn. In a corporate C-level role, you are often consumed by management overhead, leaving little time to “touch the metal.” Now, as a freelancer, I have the opportunity to experiment rapidly.
For me, this meant returning to my roots: software development.
I spent the early years of my career as a professional developer and, frankly, I missed it. However, like many managers who have climbed the ladder, I hesitated. I worried I was too rusty, that the world of coding had moved on without me.
It turns out, I picked the perfect time to return. What I lack in recent syntax memory is now completely compensated by the latest wave of AI coding agents. I have tried a couple of them and I have settled on using Claude Code for now.
But this isn’t just about me having fun writing code again. It has given me a couple of insights about the state of software engineering. Realizations that are crucial for every CEO and Tech Leader to truly understand right now.
The New Rules of Development
Here is an excerpt from what I learned returning to the IDE as a developer-turned-CTO, and what it means for your engineering organization:
- The Velocity Trap: You can now create features fast, often by a factor of 10. But this is a dangerous trap. Just because code is easy to generate doesn’t mean it should be written. Every line of code is a liability that requires maintenance. This is where experience becomes vital: knowing what to build is now infinitely more valuable than knowing how to build it.
- Structure is King: I found myself valuing statically typed languages (like TypeScript) and strict linters more than ever. The downsides of these tools (verbosity) are irrelevant when an AI writes the scaffolding for you. The structure prevents the AI from hallucinating bugs (at least to some extent), serving as the guardrails for high-speed development. It is an additional safety-net, one that I have learned to appreciate.
- The “Legacy Code” Myth: Many teams believe their codebase is too complex for AI. I challenge that. AI models don’t need to digest your entire repository at once. With just a little guidance, they can navigate complex architectures, identify relevant context, and execute changes in 30 seconds that might take even an experienced engineer a week.
- The “Rule of Three” beats “DRY“: The “Don’t Repeat Yourself” (DRY) principle is harder to enforce with AI. AI agents tend to copy first. While you can force refactoring, I’ve learned that sometimes it’s better to accept some repetition for the sake of speed, only abstracting when a pattern emerges for the third time. This is actually not a new rule in the age of AI – it has been known for more than 20 years – but is more relevant than ever with the capabilities (and weaknesses) of AI agents.
- The Death of the Static Mockup: The biggest acceleration isn’t just in writing production code, but in how we define value. In the past, we spent weeks building static mockups in Figma, hoping stakeholders understood the flow. Now, I can “vibe-code” a fully functional prototype in an afternoon. This allows us to skip the abstract discussions and gather customer feedback on working software faster and better than ever before. If your product discovery process still relies on polished, static wireframes, you are moving too slowly.
- The “Coding Speed” Fallacy: However, it is a mistake to think AI is just a developer tool. If you are not rethinking your processes around Quality Assurance, Requirements Engineering, and how your POs work, you are missing the point. Accelerating code generation without leveraging AI to speed up the rest of the chain just shifts the bottleneck downstream. There is so much more that slows down modern R&D than just typing code, and true velocity only comes when you apply AI to the entire value stream.
The 90/10 Shift
This experience brought a quote to mind that has been circulating in the tech community recently:
90% of the knowledge I built over the years is now worthless. The remaining 10% is now worth ten times as much.
I fully agree.
The 90% syntax, boilerplate, memorizing libraries is being commoditized. But the remaining 10%, among them knowing what to build in the first place, creating the right architecture, building and enforcing UX guidelines, knowing when to copy vs. abstract, strictly managing scope to name just a few of them, is what separates a successful product from a mess of AI-generated spaghetti code.
The “Zeitenwende” (Turning Point)
Many companies I consult with are struggling to adapt to this new reality.
In Germany, we have a word that describes this moment perfectly: die Zeitenwende, a drastic, historical change of times. A paradigm shift.
As a Fractional CTO and advisor, I see two dangerous reactions:
- The Developer’s Denial: Thinking AI is just a trend that will blow over.
- The Company’s Inertia: Believing you can continue with the same processes and velocity you’ve always had.
If you are a tech company feeling overwhelmed, do not push this aside. Somewhere, a startup is looking to disrupt your market. They are using AI not just for coding, but for requirements engineering, design, and QA. They are moving ten times faster than you.
Adapting to this requires more than just buying Copilot licenses; it requires a fundamental shift in how you manage engineering, how you view “seniority,” and how you execute product strategy.
I’ll close with a quote often attributed to Darwin. While likely misattributed, this quote defines our current phase perfectly:
“It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is most adaptable to change.”
It is up to you what to make of it.
