Enterprise AI Readiness Assessment
Twenty-five questions across five dimensions. Takes about six minutes and returns a scored maturity read, your binding constraint, and where to spend the next quarter. No email required to see the result.
Data readiness
Whether the data an AI system would need is accessible, documented, permissioned and of known quality.
1.Can you produce a current inventory of the data assets relevant to your top three AI use cases, including owner and location?
2.Is data quality measured against defined thresholds, with someone accountable when it drops below them?
3.Are access permissions on that data documented and enforced at a granularity you could defend to an auditor?
4.Do you know the provenance and permitted use of the data, including whether contracts or consent restrict AI training or inference?
5.Can an approved engineer get working access to a representative dataset in under two weeks?
Scoring rule: if you cannot name the document, the owner, or the last time it was reviewed, it is not a "True". Most organizations inflate this assessment by roughly one full point per dimension on the first pass.
Answer all five to continue.
This is the same instrument as Chapter 1 of the Enterprise AI Readiness Playbook. It is a structured self-assessment, not an audit, and the score is only as honest as the answers. The most useful way to run it is twice: once with the team that would build the system and once with the team that owns the underlying data or process. The gap between those two scores is usually more informative than either number.
About this assessment
What does an AI readiness assessment measure?
Five dimensions: data readiness, technical infrastructure, talent and capability, governance and risk, and culture and adoption. Data readiness and governance are the two that most often stop a programme cold, because unlike infrastructure they cannot be bought quickly.
Why does the lowest dimension matter more than the average?
Because practical maturity is closer to your weakest dimension than to your mean. An organization scoring well on infrastructure and poorly on governance is not mid-maturity, it is an organization about to build something it cannot deploy. Any dimension scoring below 8 out of 20 should be treated as a hard constraint on what you attempt this year.
How honest are these assessments usually?
Most organizations inflate the first pass by roughly one full point per dimension. The corrective is a strict scoring rule: if you cannot name the document, the owner, or the last time it was reviewed, it does not count as true.
What is the best way to run it?
Twice. Once with the team that would build the system and once with the team that owns the underlying data or process. The gap between those two scores is usually the single most informative output, because it shows where the builders' assumptions and the owners' reality diverge.
How often should we reassess?
Every six months, with the same people where possible, keeping the previous scores. The trajectory tells you more than the absolute number. A dimension that has not moved in twelve months is not a gap, it is a decision nobody has made.
What should we do with a low score?
Treat it as scoping information rather than a verdict. A low score does not mean do nothing, it means match ambition to foundations: narrow internal pilots with no customer exposure while the binding constraint is addressed, rather than a production deployment that will consume budget and produce a cautionary tale.