interesting real data on AI: We saw a company-level 3x lines of code committed since the first set of agentic coding tools were introduced. Some observations: - 3x lines is not 3x productivity/features shipped/bugs fixed/etc. It could be higher or lower in the real world since: - LLMs tends to have a higher LoC/value delivered - this only captures coding, not acceleration in other aspects of engineering (like debugging, etc.) - nowhere close to the vibe-numbers of 10x productivity #ai