KNOWLEDGE BASE · GUIDE
AI Readiness Assessment: 10 Questions Before You Spend
You are ready for AI when four things are true: you own the data, the people doing the work accept the change, an approval path exists, and one named person answers for the result. Score all four before you spend anything — ten questions, about ten minutes.
The 10 questions
Data you own
- Can someone produce two consecutive months of clean numbers for the work today? (If not, any AI return is an opinion, not a figure.)
- Do the inputs already exist in your systems — timesheets, service logs, sales exports, inventories?
People ready to change - Can you name the person whose week changes if this works?
- Will they accept the changed workflow tomorrow, not in principle, this week?
Approval path - Who signs off the spend, and what evidence do they require?
- Is the case built in their format — payback period, scenarios, worst case stated plainly?
- What happens to the money if the result disappoints?
Sponsor - Who owns the number this initiative is judged on, month after month?
- Is there a monthly check date, or does everything wait for the quarterly review?
- If the sponsor changes jobs, does the initiative survive?
Score it
| Factor | What a 0 looks like | What a 100 looks like |
|---|---|---|
| Data | No two-month baseline exists | Two months of clean CSV, ready |
| People | Nobody's week visibly changes | A named person's week changes day one |
| Approval | No evidence format agreed | CFO-format case ready before asking |
| Sponsor | No named owner | One name owns the monthly number |
Scan's readiness map implements this table: ten questions score into a 0-100 readiness index, name the factor to fix first, and recommend one first move — about ten minutes, no calls (skillpilotadvisory.ai/scan, verified 11 Oct 2026).
Sources and context
- Most AI programmes fail the board test — spend rises while cost, attrition, revenue and cycle time barely move (skillpilotadvisory.ai/about, live 11 Oct 2026).
- Malaysian enterprises often sequence AI top-down before defining the problem, which turns adoption "into an unnecessary expense" — Manminder Kaur Dhillon, CEO, Supernewsroom.ai, at World AI Show Malaysia 2026 (The Rakyat Post, 14 Sep 2026).
- Malaysia's National Guidelines on AI Governance and Ethics (MOSTI, published Nov 2024) make human oversight, accountability and transparency explicit expectations for AI use — the same factors this checklist scores.
- Malaysia established AI Malaysia on 28 July 2026 to coordinate the national AI agenda under the National AI Action Plan 2026-2030 ("AI Nation 2030") (Marketing-interactive, fetched 11 Oct 2026) — adoption pressure is rising faster than decision discipline.
From score to spend
Once the index exists: build the business case for one initiative (hours x loaded cost, 12-month model, three scenarios, payback), then prove it monthly with two months of CSV and the one sentence your boss needs. Scan runs all three in one login; 365 days of access (skillpilotadvisory.ai/scan, fetched 11 Oct 2026). For teams who want a human to drive the decision: the advisory founding sprint ends in a fund/fix/stop call (skillpilotadvisory.ai/advisory, fetched 10 Oct 2026).
FAQ
How long does an AI readiness assessment take? About ten minutes for the map itself; the honesty is in finding the real numbers, not in the form.
What score means ready? There is no pass mark. The index names the weakest factor — fix that first, buying more AI does not move it.
Is this a vendor sales tool? It is a scoring rubric against your own answers; the page states its limits and what it does not audit.
Do we need a data team? No. Two months of clean CSV plus ten answers covers the whole assessment.