Generated by Codex with GPT-5

What happened

The Pragmatic Engineer surfaced this piece in its April 7, 2026 issue, and the original post is Cycles of disruption in the tech industry: with software pioneers Kent Beck & Martin Fowler.

Gergely Orosz summarizes a live conversation with Kent Beck and Martin Fowler about how AI compares with earlier technology shifts like the internet, object-oriented programming, and Agile. Their broad argument is that AI feels bigger, faster, and messier than those earlier transitions, and that it is already changing how software teams work, how companies measure performance, and what good engineering discipline looks like.

The simple version

This piece is really about one idea: AI is not replacing the need for software craftsmanship, it is raising the cost of bad habits.

When agents can generate code quickly, weak tests, fuzzy ownership, shallow metrics, and unclear boundaries become much more painful. The flip side is that teams with good engineering hygiene can move faster than before without losing control.

Why it matters

  • Beck and Fowler argue that AI adoption is happening much faster than past platform shifts, so companies are making organizational decisions before norms have settled.
  • They call out a familiar trap: leaders measuring visible activity such as pull request volume instead of outcomes, which gets even riskier when AI makes it easy to produce lots of code.
  • The piece makes the case that TDD and strong test suites matter more in an agent-heavy world because they give teams a way to check machine-generated work instead of trusting it blindly.
  • It also highlights the human side: engineers need boundaries so they do not drift into “negative value” work where they keep pushing with AI tools after judgment and focus have already degraded.

Takeaway

The lasting message is that AI changes the tempo of software work more than the fundamentals.

The engineers and teams that benefit most will probably be the ones that keep quality bars high, stay focused on outcomes, and use agents to take on more ambitious work instead of just producing more output.