Dan Luu argues that lower implementation costs from AI make both large projects and small personal tools worth attempting when they previously did not justify the time. He describes useful tools built in a few hours while distinguishing something that works for him from production readiness. A PostgreSQL bug report submitted by a related database project contains a concrete reproducer, but that does not establish the compatibility or reliability of its entire replacement implementation. The argument pairs greater ambition with continued expertise in test design and verification.