|
Today on Applied Intelligence, we explore how one CEO curbed AI coding skepticism at his company by personally putting prompt-based development into practice. Most AI mandates flounder in the gap between executive proclamation and actual adoption. Senior engineers in particular tend to view AI coding tools with skepticism. When Warp’s CEO Zach Lloyd noticed this exact pattern taking hold at his own devtool company, he knew something had to change. “We are literally building an AI devtool at Warp and I still find it hard to get folks to change their habits by using these tools more,” Lloyd says. (Warp’s product has been undergoing a broader transformation from reimagining the terminal to building an “agentic development environment,” or ADE.) So he rolled out a pointed new policy: Every coding task needs to start with a prompt. No exceptions. Not even for him. Lloyd doesn't just talk about coding by prompt. He walks the walk by demonstrating how to do it at All Hands, in customer calls, on YouTube livestreams, and even in a custom demo he recorded just for this piece. As a former principal engineer at Google who still ships code weekly, his personal transformation carries weight. “If I’m asking all the engineers on my team to change how they’re working, it helps them to see that I really believe it — and that I can actually do it. Otherwise, I’m just another CEO who wants them to follow the AI fad,” he says. In this week’s Applied Intelligence essay, Lloyd gets into much more detail about how the mandate works and what he’s personally learned from only coding by prompt, including: - The 10-minute rule that strikes the balance of letting devs bail out if it’s wasting time, but still pushing them to try
- Why “outcome-based prompting” is outlawed
- How to ensure accountability and code quality
- Specific examples of features and fixes he’s personally built by coding by prompt
- His takes on how the job of an engineer is evolving
Thanks, as always, for reading and sharing!
-The Review Editors
|
