AI Powered Experiences — Beyond the hype
January 20, 2026
AI-Run Operations — Catching Issues Before Customers Do
January 25, 2026
This is part of the blog series, “A practical AI strategy framework — Beyond the hype”.
In my earlier post, A Practical AI Strategy Framework — Beyond the Hype, I laid out a simple idea: AI strategy only matters if it changes how work actually gets done. Not demos. Not pilots. Real systems, real workflows, real outcomes. This post zooms into the second layer of that framework: AI-built systems.
When people hear “AI-built systems,” they often assume AI is writing all the code. That’s not what’s actually happening — and it’s not the point. AI-built systems are about using AI to help teams ship faster and better. Better code quality. Fewer bugs. Less time stuck in debugging loops. Less friction in day-to-day development. The outcome is not “more AI,” it’s more momentum.
We’re already seeing this shift in the open. Both Microsoft and Google have claimed that more than a quarter of their code is written by AI. Tools like GitHub Copilot, Cursor, and Claude Code are now part of the default workflow for many developers. Vibe-coding and scaffolding tools like Vercel v0 and Lovable are becoming commonplace.
What started as code completion has quietly evolved into something much bigger. These tools can now understand entire repositories, refactor existing code, generate tests, review pull requests, and help with documentation. The scope has shifted from “write this function” to “help me manage this system.”
And yet, in most companies, the efficiency gains are hard to pin down.
Unlike big tech, many teams don’t have clean abstractions, consistent conventions, or mature development processes. Dropping powerful AI tools into that environment is like putting a high-performance car on a road full of potholes. The engine is impressive. The ride is still rough.
Improving the road — refactoring legacy code, tightening workflows, setting standards — is tedious work. But there’s no real alternative. Your competitors are already doing it, and standing still is not neutral anymore.
This is where I have a strong point of view. Humans are not disappearing from these workflows. Their role is changing. Developers are shifting from writing every line of code to orchestrating, reviewing, and approving the work that AI produces. The center of gravity moves from execution to judgment. From typing to deciding. The same thing is happening in operations — not writing every playbook from scratch, but reviewing them, refining them, and deciding when they apply.
This isn’t good or bad. It just is. And it requires a mindset shift.
When that shift is handled well, teams see real gains — often 20–40% faster delivery. When it isn’t, AI gets treated like magic. Expectations inflate, workflows don’t change, and the ROI never materializes. The most common mistake I see is assuming productivity will improve automatically. What actually happens is uneven adoption. Some developers trust the tools and move faster. Others don’t. Teams end up with inconsistent quality, inconsistent velocity, and growing frustration.
The teams that see real ROI tend to do three things consistently.
First, they set clear standards for how AI is used — where it’s encouraged, where it’s restricted, and what “good” looks like.
Second, they invest in helping teams learn how to orchestrate, validate, and govern AI-generated work.
And third, they measure real impact, not just whether a tool is enabled or licensed.
One last point that often gets missed. AI is excellent at repetitive, boilerplate work. It’s much less reliable at complex architectural decisions or deeply understanding your business context. The developers who win use AI to remove the tedious parts — and apply their judgment where it matters most.
AI-built systems aren’t replacing humans. They’re reshaping the work and redefining roles. The teams that accept this reality will ship faster — and with more confidence.
In my next post, I’ll look at AI-run operations — where AI quietly takes on the behind-the-scenes work that keeps your business running, stable, and scalable. Happy exploring — and may your AI-built systems ship faster, rand your copilots remain firmly under human command 🖖
