The problem
Teams using Claude, ChatGPT, Gemini, Copilot and Cursor have no consolidated view of what those tools actually cost, not just in dollars, but in tokens consumed, energy drawn, water used by datacenter cooling, and how that maps to the team's broader sustainability reporting.
Every vendor exposes some of this. None of them expose the full picture. Manual logging breaks within a week.
Our approach
A unified dashboard that ingests usage from each platform's API, normalises it into a common schema (tokens · electricity · water · cost), and exposes views per user, per team, per provider, and per project.
Built around three principles: provider-agnostic data model so adding a new vendor is hours, not weeks; defensible methodology so the numbers stand up to a sustainability auditor; simple export because the report has to land in an existing ESG process, not a new one.
What we shipped
- Multi-provider ingestion (Claude, OpenAI, Google, Copilot)
- Normalised metrics: tokens, CO₂ equivalent, water, cost
- Dashboard with user / team / provider / time-range views
- Export views for sustainability reporting
- Unit tests on the methodology, every number is auditable
Stack
Why it matters
For organisations that publish ESG reports, "we use AI" is starting to be a line item. This tool turns that line item into a defensible number, with sources, instead of a footnote saying "we estimate".