Did your AI energy advice lower the bill, or did the weather?

A smaller heating bill would be welcome. It would not, by itself, persuade me to keep every change an AI assistant suggested. Imagine changing the heating schedule just before a mild month: the bill falls, but you still do not know how much the new schedule helped. ENERGY STAR's existing Climate and Weather reference explains why it adjusts energy use for weather when comparing a building with itself over time. That adjustment is not proof that a particular change caused the saving. If I asked an assistant to compare my bills, I would first have it show the energy used, dates covered and price changes separately. Then ask what weather data supports the comparison, and tell it about weeks away or someone newly working from home. Without enough information, ‘the bill was lower’ is an honest answer; ‘my advice saved you this much’ may not be. This is a proposed check, not a household trial or a claim that ENERGY STAR has tested an AI assistant. The comfort matters too. If the house was colder than you wanted, record that beside the saving before accepting the schedule as a success. What would you need to see before keeping an AI-suggested heating change for the winter?