Talk about your next project
Zed executive brief

The lesson

Define what good work looks like first, then test the technology against that definition so your verdict follows from it.

The move

The essay opens with the author's experience interviewing software engineers, then under 'The Software Engineering Loop' lists the steps effective engineers repeat: build a mental model of the requirements, write code, build a mental model of what the code does, and resolve the differences. 'How about LLMs?' grants that LLMs are good at writing code, then says 'what they cannot do is maintain clear mental models.' 'But soon, right?' takes on the obvious objection and names context omission, recency bias and hallucination as causes. 'So, what now?' ends with Zed's stance that 'you are in the driver's seat, and the LLM is just another tool to reach for.'

Why it works

  • Our readSetting out the engineering loop before LLMs come up gives the reader a yardstick they can check the claim against themselves.
  • Our readConceding what LLMs do well before stating the limit makes the argument read as fair rather than dismissive to readers who already use these tools.
  • Our readQuestion-style headings like 'But soon, right?' answer the reader's objections in the order they would raise them.

What not to copy

Don't copy the mental-model thesis or Zed's people-and-agents stance, which belong to a code editor arguing about AI coding. Keep the order: define the work, grant the opposing case its strengths, name the gap, answer 'won't this change soon?', then state where your company stands.

More from Zed

More executive briefs

Steal it for your brand.

Tell us what you want to make. We’ll talk through how to put these ideas to work for your brand.