Ask an AI tool for a brand mission, vision and values and it can return something polished enough to survive a first glance.
The polish can obscure the fact that the client still has no clearer commercial decision to make.
That is not a failure of fluency. It is a failure to distinguish generating possibilities from choosing a direction.
Strategy begins before the prompt
A strategic recommendation depends on context: what the organisation is trying to change, who it needs to serve, what it can credibly deliver, which constraints matter and what evidence supports the choice.
An AI tool can help organise material, surface alternatives, compare language or expose a question you have not asked. It does not automatically know which context is authoritative, which tension matters commercially or which trade-off the business is willing to make.
If the prompt contains a vague problem, the output will often return a plausible version of the vagueness.
Generated options still need a test
The useful question is not whether an option sounds strategic. It is whether the option survives contact with the business.
Test each generated possibility against:
Context: Which real situation does this describe, and what source supports it?
Audience: Who needs to recognise or use the decision?
Distinctiveness: What would become less interchangeable if this direction were chosen?
Behaviour: What should the organisation do differently because of it?
Evidence: What would make the recommendation more credible, and what remains an assumption?
Consequence: What would the business have to stop, start or resource?
These questions turn a list of polished options into material that can be evaluated. They also make it easier to reject a good-sounding sentence without pretending that the tool has failed.
Selection is the strategic work
The strategic act is not asking for more ideas. It is selecting one direction for a reason, making the conditions visible and accepting what the choice excludes.
That selection may involve a client conversation, a claims check, a comparison with competitors, a review of delivery capability or a decision about which audience matters first. The tool can assist with preparation. It cannot take responsibility for the choice.
This is why generic polish can become more dangerous than an obviously weak draft. Weak language invites inspection. Fluent language can make an unsupported assumption feel settled.
Keep the human judgement legible
For AI-assisted strategy, retain the source material, prompt purpose, important alternatives, evaluation criteria, selected direction and unresolved questions. A future reader should be able to see how the recommendation was formed and challenge it without treating the generated text as an authority.
Use the tool to widen the field, compress repetitive material, test wording and reveal gaps. With client material, check that the tool’s data route and retention fit the engagement and that the client has authorised the use. Use human judgement to decide what is true enough, useful enough and responsible enough to carry into the work.
A simple rejection test
Suppose an AI tool proposes that a specialist business should “empower organisations through innovative solutions”. The sentence is fluent, but it gives the team no useful choice. Which organisation? What situation? What changes in the work? What can the business credibly do that makes the claim more than category language?
The strategist can reject the line without rejecting the tool. Return to the source material, name the decision the message must support and ask what the organisation would need to do differently if the direction were adopted.
AI can contribute a useful insight when it surfaces a pattern or question the team had not seen. That contribution still needs to be tested against context and evidence; the strategist makes the choice and carries its consequences.
No published comments yet.