Why does AI make transformation harder to manage?

AI deploys agents that act across your systems, creating second, third and fourth order effects that are hard to see. Andy Haley says that makes it harder to understand and influence the variables behind an outcome. He also worries about governing agent mistakes and about opaque, volatile token costs.

Should a company appoint a chief AI officer?

Andy Haley sees a chief AI officer as a shortcut that rarely works on its own. He argues the people lead and the technology lead must instead work hand in glove, because AI is merging their roles. Without a boardroom that knows how to change, the chief AI officer job is phenomenally difficult.

What is the AI maturity model of activate, prove and transform?

It is a three-stage model in which AI work moves from activation through proof to transformation. The stages overlap rather than run in sequence. Haley says value follows a sigmoid curve and only arrives at scale once skills, culture, accountabilities and a talent strategy are in place.

AI readiness assessment: Sullivan & Stanley’s three-question test

Q: What is Sullivan & Stanley's AI readiness assessment?

It is a test of three questions that a business must answer mostly yes before starting an AI programme. Are your outcomes coherent and understood? Do you understand the variables that deliver them? Can you influence those variables? Andy Haley says he will not let a client start until it passes.

Downloads | AI case studies

Sullivan & Stanley’s AI readiness assessment comes down to three questions a business must answer mostly yes before any AI programme starts. Andy Haley, CEO of the transformation firm, set them out at aiblLIVE London 2026.

He added a second rule: start the talent strategy on day one, because people take the longest to pay back. Here is how the method works.

Sullivan & Stanley’s method in numbers

3 questions to answer mostly yes before an AI programme starts.
About a year for a talent strategy to bed in, so it must run alongside the pilots.
3 overlapping stages of AI maturity: activate, prove and transform.
4 intelligences to work together: human, AI, technology and execution.

All figures as stated on stage by Andy Haley.

What are the three AI readiness questions?

Are the outcomes you care about coherent and understood? Do you understand the variables involved in delivering them? Can you influence those variables? Haley will not let a client start until it can answer mostly yes. The variables span culture, skills, data, infrastructure, partnerships, ways of working and geography.

Is half of your change portfolio actively trying to kill the other half?

Andy Haley, CEO, Sullivan & Stanley

How does AI change the readiness test?

AI makes the first question easier and the other two harder. It helps a business understand its outcomes. But agents deployed across the infrastructure create second, third and fourth order effects that leaders struggle to track. Influencing those effects is harder still, which is why governance and talent planning matter early.

When should a business start its AI talent strategy?

On day one, alongside the first pilots. Haley says a talent strategy takes about a year to bed in. The people lead and the technology lead must work hand in glove, because AI is turning their jobs into much the same job. He sees a single chief AI officer as a cop-out.

Why does human judgment matter as much as AI fluency?

Because fluency without judgment leaves people exposed. Haley maps staff on a two-by-two of human judgment against AI fluency. He warns that AI can cut out people with low judgment and high fluency. He wants deliberate training in curiosity and critical thinking, so people question flawed AI output.

So the question becomes, how do I make someone an authority to be augmented?

Andy Haley, CEO, Sullivan & Stanley

Key takeaways

Run the three-question test. Outcomes, variables, influence, before you start.
Check your change portfolio. Look for programmes working against each other.
Start talent work now. It takes about a year to pay back.
Build judgment, not just fluency. Train people to ask why.

About the speaker

Andy Haley is CEO of Sullivan & Stanley, a boutique firm working on enterprise change and transformation. He calls himself a recovering enterprise architect after about 20 years in big consulting. He spoke on the AI Infrastructure track at aiblLIVE London on 20 October 2026.

What’s in the full case study

If your board keeps asking what you are doing about AI, start with this test. The full PDF goes deeper than this page, in the speaker’s own words.

The three readiness questions and what a mostly yes looks like.
The activate, prove, transform model, and why value arrives late.
The judgment and fluency matrix for placing and moving your people.
Five actions to take back to your own organisation.

For CEOs, COOs, CHROs and technology leaders planning AI transformation. Five pages, about a seven-minute read.