An AI inventory answers a plain question: what AI does this organisation actually use? It covers projects the organisation built, AI that arrives inside products it buys, agents that act on its behalf, and tools staff use without approval.
The inventory matters because nothing else in AI risk management can be sized without it. Controls are meant to be proportionate to how material each use is, and materiality can only be assessed for a use that has been found. The Monetary Authority of Singapore’s Guidelines on AI Risk Management, issued on 7 October 2026, expect financial institutions to keep an inventory at an appropriate level of detail.
A useful inventory is built from more than one route: the project portfolio, procurement records and a declaration from staff. It is never finished, so it needs an owner and a review date.