In short
Build a simple inventory with four kinds of entry: AI tools staff use directly, AI features built into software you already have, automated workflows and agents, and suppliers whose services run AI for you. For each, note the owner, how often it runs, the model or provider if known and what data goes in. The inventory is the starting point for cutting waste, managing risk and reporting.
Steps
- Ask each team which AI tools they use and for what, including free and personal accounts.
- Check software licences and admin settings for AI features switched on by default.
- List automated workflows, scripts and agents that call AI models, with how often they run.
- Ask key suppliers whether their services use AI on your data.
- Record an owner, volume, provider and data type for each entry and review it every six months.
You can't manage AI you can't see, and much of it is switched on in tools you already pay for. An inventory turns "we use some AI" into a list you can act on: which uses are heavy, which are pointless and which carry risk.
What goes in the inventory?
| Kind of AI use | Examples | Where to look |
|---|---|---|
| Tools staff use directly | Chat assistants, image generators, coding assistants | Ask teams; check expense claims |
| Built-in features | Writing aids, meeting summaries, search answers in office software | Admin consoles and licence lists |
| Automations and agents | Scripts, workflows and agents calling AI APIs | Engineering teams; API billing |
| Supplier services | CRM, helpdesk or HR tools using AI on your data | Supplier questionnaires and contracts |
For each entry record the owner, what it is for, roughly how often it runs, the model or provider if known, and what data goes in.
What makes the biggest difference
Volume. A feature that runs automatically on every email or meeting can outweigh many people's occasional chats. Agents are worth singling out: Anthropic reported that its agents typically used about four times more tokens than chat interactions, and multi-agent systems about fifteen times more.
What next?
- Switch off what nobody uses; see switch off unneeded AI features.
- Start tracking the high-volume uses; see track AI energy per task.
- If you run an AI management system under ISO/IEC 42001, the same inventory can feed it.
Common mistakes
- Only listing tools the IT team bought, and missing personal accounts.
- Forgetting scheduled jobs that run when nobody is watching.
- Building it once and never updating it.
See our guide for business.
Sources
Related
How do I make my company's AI use more sustainable?
A practical starting plan for making a company's AI use greener: map use, right-size models, set defaults, ask suppliers and measure per task.
How do I switch off unneeded AI features across my organisation?
How IT teams can find and switch off AI features nobody needs across office, collaboration and business software, without blocking useful AI.
How do I track AI energy per task rather than in total?
Pick a functional unit such as per document or per successful task, log energy and quality against it, and report it next to the total.
Green AI for business
A practical guide for businesses on cutting the energy, cost and carbon of AI: model choice, suppliers, measurement, procurement and honest claims.