How to Start AI Coding If You Haven’t Yet

When a parent at his daughter’s school told AI Daily Brief host NLW that he had multiple AI subscriptions but had never tried AI coding, the conversation crystallized a pattern. This parent was living the "co-work life"—using AI as an assistant but never venturing into the territory where you build your own software. NLW argues that this gap is now costly. OpenAI’s enterprise research showed that by May 2025, the share of API tokens consumed agentically had surpassed those used through simple chat interfaces, and the gap is still widening. Firms in the top 10% of AI usage consumed 8.3 times as much AI as average firms—up from a 2.6x gap in January. And while engineering-related coding use grew 5x from a February baseline, finance and accounting grew 20x, sales 41x, and legal a staggering 108x. These non-engineers are not trying to become software engineers. They are using code to do their existing jobs better. The episode is a practical guide for knowledge workers who have not yet crossed that line.

Three Build Patterns: What Kind of Software Are You Making?

Before you write a line of code, NLW says you need to understand what relationship your software will have to work that already exists. He proposes three build patterns.