connecteddale

Strategy Coach = Clarity + Alignment

Generative AI in Business Strategy

Using generative AI in strategy work means putting large language and analysis models to work drafting, stress-testing and speeding up strategic thinking, not replacing the judgement call at the end of it.

A short chain of steps shows AI drafting and testing ideas, with a human decision sitting at the very end.

1 Pick a bounded task 2 Feed real context 3 Generate multiple options 4 Stress-test against reality 5 Decide and own the call
A plain sequence for using the tool without handing over the decision.

Reach for this when…

How to run it

  1. Pick a bounded task: drafting, scenario generation, summarising research, not the whole strategy.
  2. Feed it real, specific context: your numbers, your market, your constraints, not generic prompts.
  3. Generate multiple options or angles rather than accepting the first output.
  4. Have someone who knows the business stress-test every output against what they know is true.
  5. Decide and own the call yourself - the tool drafted, it didn't decide.

A worked example

Situation. Sophie Peeters ran FlitsPay, a small mobile-payments fintech in Ghent, Belgium, competing against three better-funded rivals for the same merchant base.

Applied. She used generative AI to draft ten different competitive-response scenarios overnight instead of the two her small team had time to think through, then spent the morning with her co-founder stress-testing each one against what they actually knew about their rivals' funding and habits.

Result. Two of the ten scenarios were nonsense and got dropped in minutes. One flagged a merchant-fee response neither founder had considered, and they moved on it three weeks before a rival tried the same thing.

1 Pick a bounded task 2 Feed real context 3 Generate multiple options 4 Stress-test against reality 5 Decide and own the call
Kesh Pay's founders caught the weak scenarios at the stress-test step, before any got acted on.

The catch

The output is only as good as the context you feed it, and it will produce a confident, well-written answer whether or not it's grounded in your actual numbers. Treat every output as a draft from a fast, tireless junior analyst who has never met your customers, and check it accordingly.

If nobody on the team could catch the model being confidently wrong, you don't have the judgement in the room to use this tool safely.