Why run a service all night for zero requests? Some services are busy in bursts, or only in office hours. But they run at full size, day and night. Often nobody set up autoscaling. Or the minimum is set to several copies. So you pay for peak capacity all the time, and it all draws power. Here is the fix you can use today. Add autoscaling that follows demand, instead of a fixed size. Use event-driven scaling, like KEDA, to take quiet services down to zero. Measure the cost of a cold start. Keep one warm copy only where speed really needs it. The Green Software Foundation suggests scaling workloads down when they are quiet. GreenCode takes this into account. GreenCode's Constitution skill can help an AI coding agent fix this waste. Then scale on events and check each pod's energy, using tools like KEDA and Kepler. GreenCode helps tackle this by mapping your running systems, including services that barely run. It rebuilds them as code, and shows which services use the energy. That was GreenCode Tech Tip number thirty-six. Follow for the next tip, and find more on greencode.ai/tech-tips.