Steve Ruiz — founder of tldraw, the whiteboard/canvas company — wrote a long X article about the wave of “AI is ruining coding” outrage videos. He takes the videos seriously (and says leaders especially should watch them), but his working theory is that the despair points at the wrong target: engineers aren’t miserable because AI coding is bad. They’re miserable because they’re stuck on projects where AI is genuinely boring and harmful, and they can’t get to projects where it would help.

The working theory

  • Old-world projects are still designed for pre-AI development: issue trackers, tickets, pull requests, review culture, CI, careful deploys off a pristine main branch. Feeding AI into that machine is miserable — 200 auto-generated PRs nobody can review, automations choking, understanding eroding as people stop reading code.
  • New-world projects take intent in (requirements, constraints, design, bugs) and ship excellent software out, automating away everything in between. There, AI tools are “extremely interesting and even joyful.”
  • Most projects in the world are old-world. The honest move is to push back on extreme AI adoption there — and to design new projects (or redesign old ones) to be genuinely AI-native.

A concrete experiment

Ruiz’s example is the most interesting part. On one current project he turned off GitHub Actions entirely — no CI, no CD — so agents could merge code directly without waiting on checks. Releases moved to a dedicated office machine on a 30-minute tick: pull latest, run the full test suite, make a staging release if green. If it fails, an agent is dispatched to fix the problem before the next tick.

He’s explicit that this would be a terrible idea on the tldraw SDK (“where our automations, social practices, and culture all demand a slower, more understood process”) — and that the point isn’t the specific setup, it’s designing the system around what AI is actually good at.

The warnings

  • Expect to start “knee deep in slop” — a good new-world project has high slop tolerance (internal tools, prototypes, enablement work).
  • “Don’t let an interesting process tempt you into giving broken software to innocent people” — the order is excellent software first, AI leverage second.
  • If half your team stops reading the code and PR reviews go to agents, that’s only safe on a project designed so those things aren’t required.
  • “Terrible time to be cynical”: the answer to AI making old-style development sadmaking isn’t to pull back — it’s to find or create projects where engineers work as experts with real problems to solve.