How to use this checklist
The 28 questions cover the seven dimensions that decide whether AI creates value in an organisation. They are written about observable facts, not opinions, so that two people answering honestly should land in the same place.
- Answer for the whole company, not your most advanced team.
- When in doubt, choose the lower answer and note what you would need to check.
- Ask two or three colleagues to answer separately. Where you disagree, you have found a blind spot worth discussing.
The "warning sign" column shows a common early-stage answer. "What good looks like" describes an organisation that is scaling AI with discipline. It is not the ceiling: the most advanced companies go further.
Strategy & ambition
How clearly AI is tied to business goals, where leadership wants it to create value, and how it is funded.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 1 | Is there a clear AI ambition tied to business goals? | General interest, nothing written | Company-wide ambition with priorities |
| 2 | Has leadership chosen the business domains where AI should create value? | Ideas discussed, no choice made | One to three priority domains with targets |
| 3 | How is AI funded? | Ad hoc, from team budgets | An annual AI budget |
| 4 | How involved are the CEO and the board? | Informed occasionally | Review AI progress several times a year |
Data foundations
Whether data is accessible, trusted and owned, and whether company knowledge is usable by AI.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 5 | Is key business data accessible in one place? | Some shared reports | A data platform for most domains |
| 6 | How much do people trust the numbers? | Trusted only by those who built them | Trusted in most domains |
| 7 | Are data owners named? | Informally, by habit | Named for most domains |
| 8 | Are documents, emails and company knowledge usable by AI? | Partly on shared drives | Most knowledge searchable |
Technology & platforms
Whether systems, tools and platforms let AI plug into the business and run at scale.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 9 | How modern are your core systems (ERP, CRM)? | Mostly legacy, some upgrades | Mostly cloud |
| 10 | Are approved AI tools available to staff? | Staff use personal tools informally | Most staff have access |
| 11 | Can AI connect to your business systems? | Only through manual exports | APIs for most systems |
| 12 | Can you deploy and run AI solutions at scale? | Each project builds its own | A common AI platform with monitoring |
People & culture
Whether people use AI every day, the organisation is building skills, and change is managed.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 13 | How many employees use AI in their daily work? | 5–15% | 40–60% |
| 14 | Do you have data and AI specialists? | We rely on vendors | A dedicated data and AI team |
| 15 | Is there an AI training programme? | Occasional webinars | Required training for key roles |
| 16 | How are AI-driven changes landed with teams? | Left to each manager | A common change approach |
Operating model
Who owns AI, how teams are organised, whether work is redesigned around AI, and what AI may do on its own.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 17 | Who owns AI initiatives? | IT, with occasional business input | Joint business and IT ownership of each initiative |
| 18 | Are processes and roles redesigned when AI is introduced? | Small tweaks | Redesigned for most deployments |
| 19 | How are AI delivery teams organised? | Ad hoc project teams | A central team plus cross-functional product teams |
| 20 | Are there rules for what AI may decide or do on its own (agents, automated decisions)? | Decided case by case | Human-in-control rules for all automated decisions |
Governance, risk & security
Whether AI is used safely, legally and securely.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 21 | Is there an AI usage policy? | Informal guidance | Published and acknowledged by staff |
| 22 | Are AI risks (bias, IP, privacy, wrong answers) reviewed before deployment? | Only if someone raises it | For most projects |
| 23 | Are AI tools and the data they use covered by security controls? | Not specifically | Mostly covered |
| 24 | Do you know which AI regulations apply to you (EU AI Act, PDPA, sector rules)? | Aware, not assessed | Assessed, actions under way |
Use cases & value
Whether AI delivers measurable results and successful pilots reach scale.
| # | Question | Warning sign | What good looks like |
|---|---|---|---|
| 25 | Is there a prioritised list of AI use cases? | Ideas floating around | A prioritised list |
| 26 | How many AI use cases run in production? | Pilots only | 4–9 |
| 27 | Is the value of AI measured? | Not really | For major use cases |
| 28 | Do successful pilots get scaled? | Pilots usually stall | Most reach production |
Score it in ten minutes
A checklist tells you where the gaps are. A scored assessment tells you which gaps matter most for your ambition, how you compare with peers, and what to do in the next 90 days.
Take the assessment See a sample report