AI Capability Audit: Questions Every CEO Should Ask

Most CEOs I work with think they know where their organisation sits on AI. They don't. When we run our 20-Point AI Readiness Scorecard, the gap between perception and reality averages 35%. That's not a rounding error. That's a strategic blind spot.

A proper AI capability audit covers five dimensions. Strategy and Governance: is your AI strategy documented, board-approved, and connected to business objectives? Or is it a slide deck nobody looks at? Data Infrastructure: what data do you actually have, what shape is it in, and can it support the AI use cases you're planning? Talent and Culture: do you have the people, and more importantly, do your people want this to work? Deployment and Value: which AI systems are live, what's the measured return, and which pilots have been sitting in limbo for 12 months? Vendor and Technology: who do you depend on, what's the lock-in risk, and are your contracts protecting you or the vendor?

Here's what separates a useful audit from a box-ticking exercise. You need to ask the uncomfortable questions. Show me the AI register. Every system, every data source, every decision it makes. Show me actual ROI against original business cases. Show me what failed and why. Show me vendor concentration risk. Can you prove Privacy Act compliance right now?

The 3 Types of AI Leader framework matters here. Believers get excited by audit findings and want to accelerate everything. Drivers want to fix the gaps immediately. Builders want to architect the solution properly. Each response has value, but knowing your type helps you avoid your blind spot. Believers skip governance. Drivers skip foundations. Builders skip speed.

The ADM Transparency deadline of 10 December 2026 makes this urgent. If your organisation can't produce a complete AI register by then, an audit today isn't optional. It's overdue.

53.4% of C-suite executives hide their AI use from colleagues. That stat alone tells you why audits matter. If your senior leaders won't admit what they're doing with AI, imagine what's happening three levels down.

Without a structured audit, you're operating on optimistic status updates from people who have incentives to make things look good. I've seen audits uncover 27 active pilots with only 3 in production after 18 months. I've seen 40% of staff using personal ChatGPT accounts, pumping company data into systems with zero controls. I've seen vendor concentration risk where 75% of AI capability depends on a single provider with no data portability clause.

ASIC's Chair put directors on notice in March 2026 for AI governance. If your board can't demonstrate oversight of AI capability and risk, you've got a duty of care problem under s180 of the Corporations Act. That's not theoretical. That's personal liability territory.

The audit also reveals competitive position. If your peers are 12 months ahead on operational AI, that's not something you discover in a market share report. By then it's too late. Regular audits give you lead indicators, not lag indicators. Every quarter you delay, the cost of closing the gap increases by roughly 20%.

An ASX-listed financial services firm ran our 20-Point Scorecard and discovered they scored 6 out of 20. Their CEO thought they'd be around 14. The gap was in data infrastructure (fragmented across 11 systems with no integration layer), governance (no AI register, no board reporting), and talent (their two AI specialists spent 80% of time on data cleanup). The audit triggered a complete reset: killed 19 of 22 pilots, doubled resources on the remaining 3, and built the data foundation first.

A mid-market manufacturer's audit revealed their "AI strategy" was actually a vendor sales pitch repackaged as an internal document. No capability assessment. No risk framework. No connection to business objectives. The CEO was a classic Believer type, excited about AI's potential but hadn't done the diagnostic work. We rebuilt from the Scorecard up.

A professional services CEO discovered through audit that competitors had deployed AI proposal automation 9 months earlier. Their win rates had dropped 4 percentage points without anyone connecting it to AI capability gaps. Deeper analysis showed that proposal turnaround times had blown out to 3x the competitor average, costing an estimated $2.1M in lost bids over that period. The audit converted a vague concern into a funded 90-day catch-up plan.