Only 2 of the 18 organizations we surveyed train their people on how to think with AI — how to reason with it, question it, direct it — rather than just which buttons to press. That's 11%.

Every single organization that trains this way is also using AI to build something genuinely new for customers or employees, not just a faster version of the old thing. Zero of the tool-training-only or figure-it-out-yourself organizations can say the same.

Training isn't a nice-to-have sitting off to the side of the AI strategy. In this data, it's the leading indicator for almost everything else.
Obviously correlation is not causation, but it isn't surprising that companies that invest in training people to think also measure their AI impact more and have more shared use-case policy, and most improtantly use AI to gets past "internal productivity tool" to something a customer would actually notice.

We asked leadership teams a simple but pointed question: Has your leadership team sat down and described what “two years from now” looks like, customer experience, cost structure, offering; assuming the technology keeps improving?

The answers split the room into three groups and here is how they compare:

That last data point is a single organization, so we're not going to pretend it's a statistical guarantee, but based on this data having a conversation that impacts decisions will 10x your odds of shipping something differentiated while no conversation at all appears to cap your odds at zero.

Here's the part that matters most for anyone building a 2026 AI roadmap: training and vision aren't competing line items on a budget. They're sequential. Training builds the muscle to reason about what AI can actually do. Vision is what happens when that reasoning gets pointed at the business, two years out, and turned into a plan the whole company can rally around.

Without both, companies are more likely to just use AI tools to get their current work done and at best make an incremental improvement to something.

Most of the leaders we talked to are still early. More than half (10 of 18) described their current AI training as "figure it out yourself." A third have a formal use-case policy. Nobody yet has full visibility into what their AI spend is actually producing — the best anyone reported was "partial visibility, a committee or pilot with early numbers."

It's the honest starting line for almost every company right now. When we talk ROI numbers, we know it does not happen over night, but also know that with a foundation in systems thinking, any company can eventually get there.