Plenty of companies assume they are “not ready for AI” because their data is messy. Others assume they are ready because they have a licence. Neither is the right test. Readiness means being able to take one specific use case to production.
The five dimensions
1. Strategy
Is there a named business problem with a number attached, and a sponsor who owns it? Without that, technical success still ends in a shrug.
2. Data
You do not need perfect data across the company. You need accessible, reasonably current data for one use case. Readiness is judged per use case, not organisation-wide.
3. Technology
Can your systems be reached through APIs? Is there a secure environment to run this in? Legacy systems are not a blocker, but they do change the plan.
4. People
Do you have an owner, subject experts with some spare time, and users willing to try it? Skills can be hired or partnered. Time and willingness cannot.
5. Governance
Are there rules on which data can go where? Who approves AI use? A short written policy beats both a free-for-all and a total ban. Staff are probably using AI tools already, so silence is itself a policy.
Reading the result
Low scores across the board: start with a sprint focused on one use case, and an internal, low-risk assistant.
Strong strategy, weak data: spend the first month on data for that use case only. Avoid a giant data programme.
Strong tech, weak strategy: stop building and find the business owner. This is the most common trap in engineering-led companies.
Strong everywhere: move to a production pilot, and think about a repeatable delivery process.
The usual surprise
Companies tend to over-invest in data and technology, and under-invest in ownership and adoption. Yet the projects that survive usually have a stubborn business owner rather than the most elegant data platform.
Take the free AI Readiness Assessment for your score, or work through the 40-Point Checklist with your team.