Test Everything Quickly

If someone told you that in four weeks you need to get a monkey to perform Shakespeare, how would you start?

Most people would spend the first three and a half weeks building a beautiful stage. Curtains, lighting, maybe even a ticketing system. And only at the very end would they turn to the truly impossible part: teaching the monkey to recite Hamlet.

The logic makes intuitive sense. Stage building is familiar, predictable. It feels like progress. But it’s also irrelevant if you can’t solve the only unsolved problem in the entire project. The risk isn’t in building a stage. The risk is in whether the monkey can speak.

My friend Jay Yang uses this story as a metaphor for overthinking and delayed confrontation of risk. 

I’ve seen leaders repeat this mistake in the early stages of AI adoption. They over-prepare. They gather endless stakeholders, polish documentation, align roadmaps, and spec out interfaces. All of which are useful eventually, but none of which answer the real question: can AI actually perform the task you need it to?

The hard part is never standing up the infrastructure or connecting systems. We know how to do that. Those are known problems. The question that matters is proof of concept. Will this model pass muster with your compliance team? Can it reliably write in your brand’s voice? Will it summarize sensitive documents without errors? Until you test those unknowns, everything else is theater.

The leaders who succeed in AI adoption do the opposite of stage-building. They pull the riskiest problem to the front. Instead of planning a six‑month rollout, they run a two‑day experiment. They sit with the monkey. They see if Shakespeare is even in play.

Here’s the framework I keep returning to when scoping AI work:

1. Identify the unknowns. Be brutally honest about what you don’t know, and where the failure points are.
2. Prototype on the edge. Mock it up, test it fast, break expectations early.
3. Scale after proof. Once you know the monkey can speak, then, and only then, worry about the stage, the lights, the curtain call.

It’s tempting to work on the comfortable parts of AI adoption first. But leaders are measured by results, not rehearsals. Don’t build a perfect stage. Get the monkey on script.

‍