What’s your strategy for exponential change?

Someone told me about a company going all-in on AI.

They started by giving everyone a Copilot license. 10,000 seats. About 3.6 million euros per year.

And trained everyone before anyone knew what to use it for.

Three months later, only a few hundred users are active, the majority of licenses unused.

The approach was: deploy the tool, run training sessions, get the benefits!

That can work for software with defined processes (ERP, CMS, CRM).

But AI is different.

AI explores what we haven’t discovered yet.

AI can redefine entire processes. Every 3 months.

How do you plan when change is exponential and impact unknown?

When uncertainty is high, the playbook changes

Here are some strategies to help you adapt to this AI hurricane, the famous AI-rricane:

• Set boundaries. Define guidelines where AI can operate safely. Freedom to experiment requires clear lines.
• Inspire, don’t instruct. Show, don’t tell. Hands-on demonstration beats training decks.
• Small bets, real autonomy. Teams closest to the problem test and try. Bottom-up, not top-down.
• Fund traction, not proposals. Back what’s working, not what looked good on a slide. Results beat projections.
• Stay honest. Let go of what isn’t working. Double down on what is.

Stay honest might be the hardest, as companies work with forecasts, roadmaps and ROI.

Saying “I don’t know exactly where this goes” becomes courage, vulnerability and honesty.

And that’s what works.

That Copilot company?

They could have started with 500 licenses across 10 teams, measured real usage, and scaled from there, instead of committing to 10,000 from day one.

This tension between planning and learning is hard to figure out alone. It’s easier with peers who face the same thing.