Is your organisation actually ready for AI? The 5 questions that cut through the hype

Every board meeting has AI on the agenda. Every leadership team has been asked to develop an AI strategy. Most of them are not ready to execute one — not because the technology isn't there, but because the foundations aren't.

Here are the five questions we ask every organisation before recommending they begin an AI project. They're not technical questions. They're operational ones. And most organisations can't answer all five.

Question 1: Is your data actually usable?

AI models are only as good as the data they learn from. Before asking "what can we build with AI?", ask: is the data we'd use for this clean, consistent, and accessible? Is it in one place, or spread across seven systems? Is there a team that owns its quality?

The most common failure mode we see: an organisation invests in an AI project, and six months in discovers that the underlying data is too inconsistent for the model to learn anything useful. The AI project becomes a data quality project in disguise — at twice the cost and half the speed.

Question 2: Do you have labelled examples of what good looks like?

Supervised learning — the most reliable form of AI for most business problems — requires examples. To build a model that detects fraudulent transactions, you need a dataset of labelled transactions: "this one was fraud, this one wasn't." To build a contract review system, you need contracts that humans have already reviewed and annotated.

How much labelled data you need varies enormously by problem. But "we'll figure out the training data later" is not a plan. The labelling effort is often the most expensive part of an AI project, and it almost always takes longer than expected.

Question 3: Who owns the output — and what happens when it's wrong?

AI systems make mistakes. A contract review system misses a clause. A recommendation engine suggests the wrong product. A fraud model flags a legitimate transaction. Who is accountable? What's the escalation path? How quickly can a human override the system?

Organisations that haven't answered these questions before deployment create problems that are much harder to solve after something goes wrong. Human-in-the-loop design isn't a concession — it's good engineering for any system where errors have consequences.

Question 4: Can you measure whether it's working?

This sounds obvious. It isn't. "The model is 94% accurate" is a metric. "Customer satisfaction has improved by 8 points since deployment" is a business outcome. The second matters; the first is only interesting if it predicts the second.

Before starting any AI project, define: what does success look like in business terms? How will you measure it? What's the baseline you're comparing against? If you can't answer these, you can't know whether the project worked — and you can't justify the next investment.

Question 5: Is leadership ready for a technology that produces probabilistic outputs?

This is the question most people find uncomfortable. Traditional software is deterministic — input A produces output B, always. AI is probabilistic — it's right most of the time, wrong sometimes, in ways that are not always predictable. Leaders who are used to software that either works or doesn't find this deeply unsettling.

The organisations that succeed with AI are the ones where leadership understands and accepts this tradeoff. Not blindly — they put in the monitoring, the human oversight, and the feedback loops. But they don't expect perfection, and they don't pull the plug the first time the model makes a mistake.

If you can answer all five questions well, you're probably ready to start. If two or three are unresolved, start with those rather than with the AI project itself. The foundation work isn't glamorous, but it's what determines whether the AI investment pays off.