The future is stuck in procurement

Recently I had two conversations in the same week. The first was with someone deep in the AI scene. He told me, with total conviction, that the next twelve months will change everything. The second was with the leadership team of a healthy, profitable company. Twenty minutes in, it became clear that their entire AI strategy was one free ChatGPT account. Shared. Which IT had just blocked.

Here is the uncomfortable truth: both of these worlds are real. AI really is moving that fast. And companies really are moving that slow. The mistake is believing the first speed has anything to do with the second.

Technology moves at the speed of research. Companies move at the speed of trust. Of budget cycles. Of job descriptions. Of the quiet fear of looking stupid in a meeting. A model can double in capability in a year. An organization cannot double its willingness to change in a year. It never has, with any technology, and it will not start now.

The twelve-month people are not lying. They are in a bubble. They talk to each other, demo to each other, and scare each other, and for them the frontier really is daily life. But the median company is nowhere near the frontier. The median company has one enthusiast, a blocked domain, and a working group that has been evaluating tools since last spring. The most disruptive technology in the building is a free account someone set up on a private email.

So try a thought experiment with me. Imagine the frontier froze tonight. No new models, no new releases, nothing, for ten years. Now ask yourself honestly: how long would it take your company to absorb just what already exists? To actually rebuild your workflows around it, train your people on it, trust it with real work?

Every leader I ask goes quiet at this point. Because the honest answer is years. The technology you already have access to, today, exceeds what your organization can digest this decade. Last year's model, in everyone's hands, with permission to use it, quietly rebuilds a company. The next five releases, watched from a distance, change nothing.

And this is the genuinely hopeful part. You are not twelve months behind OpenAI. You are two decisions behind yourself. Decision one: give people real accounts and real permission, in writing, instead of a policy vacuum where everyone uses it secretly anyway. Decision two: pick three workflows that hurt, and rebuild them properly, with the people who do the work in the room.

That's it. You do not need a strategy for the future of intelligence. You need a decision about Tuesday.

So let the bubble talk about the next twelve months. The companies that win this decade will be the ones that finally used the last twelve.

Go unblock the domain.