Responsible AI
Your AI use, repaid in trees.
We count the work our models do per subscription, convert it to CO₂ and plant trees for it every quarter through Tree Nation.
Why the footprint is small
Claire is frugal to begin with.
Three reasons the footprint is limited, before a single tree is planted.
Short conversations
A question to Claire takes a handful of interactions, not an hours-long session. The arithmetic happens in your administration, not in the model.
Light models
Most tasks run on a light, efficient model. We only bring in a heavier model when a task calls for it.
No model training
We do not train models of our own. We use existing models, so our use adds no training run.
How we calculate
From prompt to tree.
Count
All model work is measured in tokens: the pieces of text the model reads and writes.
Convert
Tokens are converted to kilowatt hours and CO₂e, based on publicly available data per model.
Report
We report every year on the CO₂ footprint of our AI use and the reduction we achieve.
Plant
Tree Nation plants the trees that cover the total every quarter. Publicly verifiable.
Accountability
A receipt every quarter.
After each quarter closes, we publish the statement: the use, the reduction achieved and the number of trees planted, with a link to the planting on Tree Nation.
No figures before they are final. The first receipt follows after the current quarter ends.
See the DataFlowr forest →DataFlowr · accountability
Q3 2026
Published after the quarter closes.
Questions about our approach?
You speak to a founder directly, not a sales team. Audit questions are welcome too.