AI in Practice

AI can generate code quickly. That part is no longer interesting.

What matters is what happens when you attempt to build a real system with it.

This was not a toy project. It was a full system, built end-to-end, with clear boundaries, real constraints, and expectations of correctness.

The goal was not to see if AI could write code. The goal was to see how it behaves when it is treated like a development team.

AI is not a shortcut around engineering.
It is a mirror of how you approach it.

The most immediate realization is that AI does not struggle with implementation. Given clear direction, it can produce large amounts of working code quickly. Services, data models, interfaces, integration points. All of it can be generated at a pace that would be unrealistic for a single engineer.

Speed is not the limiting factor.

Clarity is.

When direction is precise, AI output aligns. When direction is vague, output diverges. Not randomly, but consistently. It will fill in gaps with reasonable assumptions, but those assumptions are not coordinated across the system.

That is where problems begin.

Early in the process, it becomes obvious that AI will not enforce consistency on its own. It will generate similar solutions in slightly different ways. Naming drifts. Patterns vary. Boundaries blur.

Individually, each piece works.

Together, they do not form a coherent system.

This forces a shift in how development is approached.

Instead of writing code directly, the primary work becomes:

• defining structure
• setting constraints
• correcting deviations
• reinforcing consistency

The role of the engineer changes from writing implementation to directing it.

This is where most of the value is created.

Without that direction, AI produces output that looks complete but behaves inconsistently under real conditions. With that direction, it becomes an effective extension of the engineer.

There is also a practical limit that becomes clear very quickly.

AI does not retain system-wide context reliably over time.

It does not remember every decision. It does not enforce previously established patterns unless they are explicitly restated or encoded into the workflow.

This means that consistency must be externalized.

• architecture must be documented
• patterns must be defined
• constraints must be repeated or enforced

Otherwise, drift is inevitable.

This is not a flaw. It is a characteristic.

Treating AI as if it “understands” the system leads to subtle errors that compound over time. Treating it as a capable but stateless contributor produces better results.

Another observation is that AI will confidently generate incorrect solutions when constraints are unclear.

Not obviously broken solutions. Plausible ones.

This is where experience matters.

You have to know when something is wrong, even when it looks right.

That is not something AI can validate for you.

It will not tell you when a boundary is misplaced. It will not challenge an assumption unless explicitly asked. It will not protect the system from poor decisions.

It will execute them.

This reinforces a consistent theme.

AI amplifies the engineer.

If the system is well-designed, AI accelerates progress. If the system is poorly defined, AI accelerates inconsistency.

The difference is not in the tool. It is in how it is used.

Over time, a stable pattern emerges.

AI works best when:

• the architecture is defined up front
• constraints are explicit
• patterns are enforced consistently
• output is reviewed with intent

Under those conditions, development becomes faster without losing coherence.

Without those conditions, development becomes faster and less predictable.

The experiment makes one thing clear.

AI is not replacing engineering.

It is raising the bar for it.

Because when implementation becomes cheap, the cost of poor decisions increases. Systems can be built faster, but they can also become unmaintainable faster.

That is the trade.

AI changes how software is built, but not what makes it good.

Clarity still matters.
Structure still matters.
Judgment still matters.

If anything, they matter more now than they did before.