Skip to content

Engineering · 7 min read · Unpublished draft

AI-assisted software engineering: how we use it, and where we don’t

How AI coding tools fit into a disciplined engineering workflow — and the review practices that keep quality high.

Draft outline

This article hasn't been written yet. Below is the planned outline. Edit it in src/data/insights.ts, then set its status to published.

Planned outline

  1. 01

    Where AI assistance helps

    • Boilerplate and scaffolding
    • Test generation as a starting point
    • Exploring unfamiliar code

  2. 02

    Where human judgment stays essential

    • Architecture decisions
    • Security-sensitive code
    • Business rules

  3. 03

    Keeping quality high

    • Review standards for AI-assisted changes
    • Client data and tool policies
All insights