As environmental professionals rely more heavily on AI to crunch sustainability metrics, model scenarios and draft impact reports, the energy used to run that analysis is becoming a measurable part of the organisation's own carbon account, and one that's often outside anyone's direct control. This workshop will explore how to take that control back, through:

• Understanding where AI's energy use actually comes from: why a single complex query can use as much power as an LED bulb left on for four hours, and how this scales across looped queries, analysts and projects
• Identifying the hidden costs of off-the-shelf AI tools: one fixed model for every task, "thinking" always switched on, and data sitting on someone else's servers
• Exploring how an agnostic AI infrastructure lets the business choose which model handles which job
• Evaluating the case for keeping data in-house and matching effort to task, as a lever for reducing both AI footprint and AI cost

Attendees will leave with a practical framework for assessing their own AI infrastructure choices and reducing the carbon and cost impact of their AI use.