NAKEDAI PRACTICAL RESOURCE

The AI Spend Sanity Check 

Four questions before AI spend becomes speculative.

 Before committing money, people or operational change, leaders should be able to explain the problem, the workflow, the data and the expected value in plain English.

​

​

Before AI spend becomes speculative​

Use this check before signing off spend where the idea looks impressive, but the problem, workflow, data or value is still not clear enough.
 
AI spend becomes speculative when it solves a problem the business has not properly named.
 
A strong business case should survive plain English questions from the CEO, COO and CFO before money, people or operational change are committed.​

​

Question one: What problem are we solving?​

​

• Avoid vague goals such as “improve efficiency” or “increase productivity”.

• Name the actual business problem.

• Ask whether the problem is expensive, frequent, risky or strategically important enough to justify an AI intervention.

• If the problem cannot be explained clearly without mentioning the proposed technology, the business case needs more work.

​

Question two: Have we fixed the workflow first?​

​

• AI will not rescue a process that no one understands.

• If the workflow is inconsistent, unclear or full of avoidable handovers, simplify it first.

• Automating confusion does not remove it. It can make the confusion faster, harder to see and more expensive to correct.

Question three: Is the data ready for this use?​

​

• Check whether the data is accurate, accessible, permissioned and appropriate for the decision being made.

• Consider who can use the data, what limitations it contains and whether it is suitable for the proposed outcome.

• Bad data does not become safe because a new system is placed on top of it.

​

Question four: What value will this create?

​

• Name the financial, operational, customer or risk outcome you expect to improve.

• Explain how that value will be measured and when the evidence will be reviewed.

• If the value cannot be measured or described clearly, approval should pause until the business case is stronger.

​

Is there a simpler alternative?

​

Before committing to AI, ask whether a simpler process or decision change should be tried first.
 
The right answer may be to remove an unnecessary step, clarify responsibility, improve the underlying data or simplify the existing workflow.
 
Choosing the simpler option is not a failure of AI strategy. It is evidence that the organisation is making a disciplined business decision.

​

​

The leadership check

​

Use these questions with your leadership team.
 
If an answer is vague, leave it blank and treat that as useful evidence.

​

​Problem statement​

What specific business problem are we solving?
 

Workflow status
Is the process clear enough to improve, or are we automating confusion?
 

Data readiness
Is the data accurate, accessible, permissioned and appropriate for this use?
 

Expected value
What value will be created, and how will we know?
 

Simpler alternative
Is there a simpler process or decision change we should try first?

​​

​

When the business case is not ready

​

If you cannot clearly name the problem, explain the workflow, assess the data and describe the expected value, the decision is not ready.
 
Take a Human Pause before approval.

​

​