Stage 2 of 8 · Set up and assess

Inventory and intake

Know every AI system and control new ones at the door.

For reference purposes only. Not legal advice.

SimplifiedGeneral guidance in plain language. Region callouts are short summaries; read the region pages and sources.Source: NIST AI Risk Management Framework · Checked 3 Oct 2026

What this stage is for

You cannot govern what you have not found. Start with a register of every AI system, including AI built into vendor products and tools staff use without approval (often called shadow AI). Then add an intake step so every new idea or purchase is triaged before it is built or bought.

Who is involved

AI governance lead, Product owner, Procurement, IT, Engineering and data science.

Key outputs: AI register, intake records.

Step-by-step tasks

Tick each task as you complete it. Your ticks stay in this browser tab unless you export your workspace.

Region callouts

Shown for the regions in My regions (header). The region pages are never hidden.

EU

A register is the base for working out each system's risk tier and your role as provider or deployer. European Union page

US

Add the states and cities where each system is used. State and city rules (for example hiring tools in New York City) depend on where people are affected. United States page

UK

Record where personal data is used. It drives data protection impact assessments. United Kingdom page

IN

Record which systems use digital personal data of people in India, and which host or generate content for users there. India page

Tools and templates

Tools

Templates

Common mistakes

  • Counting only models your team built and missing vendor AI.
  • A register that is out of date a month after it is created.
  • Intake that is so heavy people route around it.

You're done when…

  • Every known AI system is on the register with an owner.
  • New uses cannot start without intake.
  • Each system has a review date.

Reviewed 3 Oct 2026.

← Stage 1: Set up governance · Stage 3: Assess risk and impact →

Was this page useful? Tell us what is missing or wrong ↗ · Report a correction