AI readiness is not an abstract maturity level. It is the answer to a practical question: for the use cases we care about, what needs to be true — and what do we need to fix? Use this checklist as a starting point; a formal AI readiness assessment goes deeper.
1. A clear business problem
Can you name the process, who does it today, how long it takes and what a better outcome looks like? If not, start with AI strategy, not tools.
2. Measurable success
Agree how value will be measured — time saved, turnaround, quality, customer response — before build, so results are demonstrable.
3. The right data, in usable form
Does the information the use case needs exist, is it reasonably accurate and current, and is it in a system AI can reach?
4. Permissions and over-sharing
AI assistants show users whatever they can already access. Files shared with "everyone", stale guest access and broad SharePoint or Drive permissions become an AI problem the day an assistant is switched on. This is usually the most important readiness task.
5. Security controls
Multi-factor authentication, managed devices, logging and data loss prevention reduce the risk of AI-assisted data leakage. See Microsoft 365 security.
6. POPIA and personal information
If a use case processes personal information, define the purpose, minimise what is used, control where it is processed and stored, and document it. See POPIA and cybersecurity.
7. Integration
Which systems must the AI read from or act on, and do they have APIs or connectors? See system integration.
8. Governance and policy
Who approves new use cases? Is there an acceptable-use policy telling staff which AI tools they may use and with what information?
9. Ownership
Each use case needs a business owner accountable for its outcome, not only an IT owner.
10. Skills and adoption
Do people know how to use the tool well and check its output? Adoption needs planning — see change management.