Featured Software Factories That Work for Humans
Your team is shipping more code than it can read.
That is the quiet arithmetic behind most AI coding programs right now, and it doesn’t resolve by hiring more reviewers or asking the existing ones to concentrate harder.
In this issue’s lead article, John Rhodes builds on a session from the AI Conference arguing that the way out is automated evidence: the pipeline proving what the code does, instead of a person certifying that they looked.
Legacy modernization teams got here first, for an unglamorous reason. They have been working on systems too large to read since long before agent existed.
Read the full article → |