Major tech companies like GitHub, Google, Dropbox, and Microsoft are measuring AI's impact on software development using a combination of traditional engineering metrics (change failure rate, PR throughput) and AI-specific metrics (time savings, adoption rates, customer satisfaction). The research reveals that 85% of engineers use AI tools at work, but measuring their true impact requires tracking both speed and quality metrics together. Companies are finding AI particularly effective for code migrations and grunt work, while being cautious about data security and long-term code maintainability. The measurement approach combines system data, periodic surveys, and experience sampling to get a complete picture of AI's effect on developer productivity.