← All writing
aiengineering

AI in the loop, not in the driver's seat

How I fold AI assistance into real engineering without giving up review, security, or taste.

AI-assisted development isn't magic, and it isn't the thing coming for your job. It's a power tool, and power tools are wonderful right up until the moment you stop respecting them.

Human-in-the-loop, always

Every change the model suggests goes through the same review I'd give a teammate's PR. Does it match the architecture? Is it secure? Does it handle the edge cases nobody thinks about until 2am? The model drafts, a human decides, and that boundary doesn't move.

Fundamentals don't get cheaper

If anything, AI makes the fundamentals worth more, not less. SOLID, clean code, semantic HTML, accessibility: that's the toolkit you use to judge what the model hands you. You can't review code you don't understand, and pretending otherwise is how bad things ship.

Where it actually shines

  • Scaffolding the boring 80% so I can spend real attention on the 20% that matters
  • Getting me up to speed on an unfamiliar API fast, then verifying every claim against the docs
  • Catching the dumb typo-class bug before a human ever has to

Used this way, it was never about writing more code. It's about spending judgment where judgment counts.

Lamadrid Labs © 2026