Skip to content
Ákos Kőműves

Newsletter

For engineers who pair with AI as a coworker, keep the thinking, and ship value instead of slop.

  1. Why do we even have loops? Supplements for engineers who skipped leg day
  2. Delete All Your Skills The essence probably fits on a pamphlet
  3. Why a handful of engineers will beat your agent fleet Motivated, capable engineers with a deep understanding of the business will always out-ship meat proxies and create a superior product.
  4. We went from human slop to AI slop. That’s fine. Hot take: AI-written code is not that bad.
  5. Agentic coding is boring if you only write code Play the metagame
  6. The skill that outlasts coding No, it’s not the developers who use AI nonsense.
  7. Time-to-market vs time-to-regret Is code still important?
  8. The Year of Side Projects If you ignore the advice of people whose success depends on AI development going vertical, you’ll find yourself in a different storyline.
  9. AI is too dumb. Let me do this! Why it’s still hard to do “AI” right
  10. The Local Agentic Future is Already Here — Just Not All at Once How the transition period will look like
  11. Why you should be excited about local LLMs too cheap models are good
  12. I Tried to One-Shot My Entire Sprint With AI and Failed Why Agentic Coding Is So Difficult in an Enterprise
  13. How to Give AI Agents Safe Access to Your Production Database This boring fix is older than agentic coding
  14. Cursor Made Agentic Coding WYSIWYG Hi Friends,
  15. Why I Cancelled My Claude Code and ChatGPT Pro Subscriptions (for Side Projects) Last week I posted a simple tweet and it exploded (300k+ views).
  16. Power Plants and Steam Engines: How SDKs Will Solve the Compute Crisis I found two technologies that fight both model lock-in and the growing infra crisis.
  17. I don't know — but here's how I'd find out Why defensiveness isn't seniority — and what to say instead.
  18. How to write coding agent teams that actually work Four design choices behind a code-review team that ships useful feedback.
  19. Nothing Failed. That’s the Problem. Your agent optimizes for green, not for truth.
  20. Your AI Tests Are Probably Useless How to write better tests with coding agents