The core question for AI adoption and transformation. Answer this one before any other.

How fast are you learning? That's it. That's the question you should be asking.

I see two basic bottlenecks to AI adoption: (lack of) creativity and access to enough compute. This last one is only a problem if you've solved the first. So, start there.

Creativity means imagining what we could do differently. Where could AI help us solve a problem, meet a customer demand faster, or try something that previously wasn’t practical?

Once those ideas start flowing, we need enough compute to explore them. An idea leads to an experiment. The experiment teaches us something. That learning sparks another idea. Compute needs to keep up with that cycle.

As of November 4, 2026, we’re all still learning. We don’t know what AI adoption will look like in the future. We have to experiment to discover what works.

That makes the speed of learning a central challenge.

I’d start tracking it with a few questions:

• How long does it take us to turn an idea into an experiment? (The goal is learning, so make the experiment around core questions you still have)
• How quickly do we get useful feedback from the market or direct users/customers?
• How often does what we learn change what we do next? The rate of change here is critical. No change, no learning.

To accelerate that learning, we need to make experiments easier to run: give people room to explore, access to compute, and a way to share what worked, and what didn’t. Keep experiments small enough to get answers quickly. This is critical, a 5 min experiment is worth more than ten 50 min experiments. That's a butter lesson many haven't learned yet.

But the key goal that we must keep focused on is learning: every experiment needs to deliver that learning. And we must keep track and share those learnings with the rest of the team.

How are you tracking your speed of learning? More importantly, how are you accelerating it?