Stephen Butler highlights the benefits of designing the right structure for using LLMs before diving into strategy.
BCG taught me strategy is about finding a narrative in a mess of facts, making coherent choices, and sticking to them.
Humans aren’t naturally great at any of this.
❌ We shoehorn facts to fit what we want to hear or do.
❌ Groupthink overwhelms coherence.
❌ Execution discipline becomes harder with time and distribution.
In this light, using an LLM as a thought partner makes intuitive sense.
✅ LLMs are by nature the world’s best pattern spotters.
✅ They don’t shy from calling out logical errors that humans may be embarrassed to raise.
✅ Once a path is chosen, they can recall and bring us back to it.
I’ve been working with an increasing number of clients to form and execute strategy using #customGPTs, #GeminiGems and #ClaudeProjects (the last being far and away the best).
With the right structure, LLM agents can be amazing—shifting the work of consultants like me away from pure strategy formulation toward data gathering, refining specifics and driving broader engagement.
But what’s the right structure?
A recent Harvard Business Review study by Natalia Levina (NYU Stern), Angelo Romasanta and Llewellyn Thomas found that LLMs are natively drawn towards what everyone hates about consultants: buzzwords and clichés they call #aitrendslop.