Skip to main content

Vibe Coding Works Best With Engineering Discipline

Core idea: AI amplifies good processes and bad habits alike.

The Promise and the Risk​

Vibe coding — using AI to rapidly generate software from natural language prompts — is incredibly powerful. It lowers the barrier to entry, accelerates prototyping, and lets non-developers build functional applications.

But here's the uncomfortable truth: AI amplifies whatever processes you already have. If you have good engineering discipline, AI makes you dramatically more productive. If you have bad habits, AI helps you ship bad code faster.

The Four Forces​

Steven Brovich identified four forces that are reshaping software teams in the agentic era. Understanding them is essential to using AI responsibly:

1. Expert Multiplier​

A group of senior, knowledgeable people with AI at their disposal can achieve an order of magnitude increase in speed (see Project Mantle). But this only works when the humans have deep expertise to evaluate and steer the AI's output.

2. The Bottleneck Shift​

The question is no longer "can we build it?" It's "do we have the data and can we decide fast enough to keep up with what we can build?" Implementation speed has outpaced decision-making speed.

3. The Verification Tax​

AI generates code 10 times faster, but it's three times harder to validate. The reviewing bottleneck eats the velocity if you're not careful. Every line of generated code needs to be reviewed, tested, and validated.

4. The Deskilling Trap​

Juniors using AI can ship about 17% more code, but they understand 17% less of what they've shipped. The people coming through the pipeline are getting faster and less grounded at the same time.

All four are true simultaneously. The leader's job — and every developer's job — is to hold the tension.

Engineering Discipline in the AI Era​

Vibe coding without discipline leads to:

  • Technical debt acceleration — bad architecture decisions made at AI speed
  • Security vulnerabilities — generated code that looks correct but has subtle flaws
  • Fake confidence — shipping code you don't fully understand
  • Maintenance nightmares — code that works today but is impossible to extend tomorrow

Vibe coding with discipline means:

Use Specifications​

Before you prompt, write down what you're building. Define the requirements, constraints, and success criteria. A good spec produces good AI output. A vague prompt produces code you'll have to rewrite.

Test Generated Code​

AI code is not exempt from testing. Unit tests, integration tests, and security scans should run on every generated change. If you can't test it, don't deploy it.

Review Architecture​

AI generates code at the function and file level. It doesn't always get the architecture right. Review the system design, not just the implementation.

Treat AI as an Accelerator, Not an Infallible Engineer​

AI is a powerful tool. It is not a replacement for engineering judgment. The human is responsible for every line of code that ships — whether they wrote it or not.

The Operating Model Matters More Than the AI​

As Brovich concluded:

"The companies that win in the next decade aren't the ones with the best AI. They're the ones with the best operating model around the AI."

The technology is advancing rapidly. The models will get better. The tools will get cheaper. But the discipline — the specifications, the testing, the review, the architecture — that's what separates successful AI-assisted development from chaos.

What This Means for You​

  • Good habits matter more than ever. AI amplifies your existing practices. Make sure they're worth amplifying.
  • Discipline is not anti-AI. It's the opposite. Discipline is what lets you use AI safely and sustainably.
  • Start with specs. Before you write a prompt, write a specification. It will save you time, money, and headaches.
  • Test everything. AI-generated code needs the same rigor as human-written code.
  • Review before you deploy. The verification tax is real. Pay it.
  • Vibe coding is not a replacement for engineering. It's an accelerator for engineers who know what they're doing.

Portions of this article are based on insights from Steven Brovich's talk "A leader's guide to advanced team structures in an agentic world" at AWS Events. You can watch the full talk here: A leader's guide to advanced team structures in an agentic world.

Last updated on