Choose the right model
Not every task needs the largest model. A smaller one that does the job needs less computing power.
Every AI request takes energy
That is why we use AI with care: With the right model, with efficient applications and with the goal of saving resources overall.
1.5%of the world’s electricity went to data centres in 2024.
2030By then, their electricity demand is projected to more than double.
AIis the biggest driver of this growth.
Source: International Energy Agency (IEA), Energy and AI, 2025
Five principles for lean AI
Develop efficiently
Only as much computing power as needed: Lean interfaces and no unnecessary calls.
AI only where it helps
Up front, we work out which tasks suit AI and which do not.
Save resources overall
Good applications save trips, paper, time spent searching and duplicate work. Only then is the energy they use worth it.
Keep checking after launch
Once an application is in use, we keep checking whether models and tools still fit.
Using less starts in our own company
Our goal: A CO2 footprint that is as small as possible.
Not a single sheet of paper
From the first call to the invoice, everything runs electronically. Nothing gets printed.
Signed without a pen
You sign proposals electronically.
The invoice arrives digitally
As an e-invoice, not on paper.
Computing power from renewable energy
The data centres of our service providers run on renewable energy.
By bus and train to meetings
We travel to meetings by public transport instead of by car.
Devices that stay in use
We use our devices for a long time instead of replacing them often.
Where could AI save resources for you?
In a first call, we work out which task is worth it and which model is enough.