with10-50 per cent typical load savings right now.

Thunder Energy’s report on AI use contains a case study in which 98 per cent of electric vehicle charging loads were shifted away from peak grid hours to off-peak hours by Octopus Energy's Intelligent Octopus. Through an app, customers set what time they want their EV charged by, then an AI optimised the timing. Total household load in peak hours also fell by 35-50 per cent. Only 1-2 per cent of customers elected to over-ride the AI, confirming its efficacy.

In another example, 13 per cent energy savings were achieved by Brainbox AI researchers ‘nudging’ the setpoint of heating systems lower, at a university building in Ottawa in winter-2025. The rationale is that most building occupants touch their thermostats less than once per month. Hence the AI sneakily nudged down the setpoint. If the occupants do not object, it nudges it down further. And so on, until the user nudges it back up again.

The use of AL, together with a little psychology, could help rewrite some assumptions about the need for more capital-intensive projects.

---