The discernment gap
The biggest AI skill gap in 2026 is not technical.
There is a question that comes up in almost every room we teach in, and it is always the wrong one.
Which AI tool should I be using?
It is a reasonable question. It has an answer. And the answer will be out of date by roughly the middle of next year, which is why it is the wrong thing to be asking.
The right question is quieter. Not which tool, but whether this is a task a machine should be touching at all.
Where it actually goes wrong
Almost nothing we see going wrong with AI inside organisations is a tool failure.
A consultant uses AI to write a condolence note to a client. The tool did what it was asked. The judgment failed before the prompt was typed.
An operations lead automates an approval process that was already producing bad approvals. The automation works perfectly. It now produces bad approvals at four times the speed.
A team pastes commercially sensitive material into a free tool because it was the one they had open. Nobody read the terms. Nobody thought to.
None of those people lacked tool knowledge. Every one of them lacked a framework for deciding.
Why nobody teaches it
Tool knowledge is easy to sell. It demonstrates well, it packages into modules, and it produces a visible result inside an hour. A person leaves a workshop able to do something they could not do that morning, and that feels like value.
Judgment does not demonstrate well. It shows up as restraint, which looks like nothing happening. It is slow to build. It cannot be assessed with a quiz.
So the market has drifted, almost entirely, toward the skill with the shortest shelf life.
What it takes to build instead
Three things, and none of them are a tool.
A framework the person can apply to a decision they have not seen before, rather than a list of approved use cases that goes stale.
Someone who can hold them to it. Judgment develops against resistance. It does not develop from watching a video.
Real work. Practice on the actual decisions in front of them, not case studies about fictional companies.
That is a harder thing to build and a harder thing to sell, which is roughly why it does not exist in most places.
What we built
The SHE IS AI Applied Intelligence Pathway opens today.
The Catalyst cohort runs six weeks from 15 September, across twelve modules. Participants leave with eight working artefacts rather than notes: an AI values statement, a personal governance framework, a responsible AI checklist, a prompt portfolio in their own voice, an integration plan, an ethical decision-making guide, a pathway map and an applied use case project.
We call that set the Catalyst Portfolio.
There is an application and a conversation rather than a checkout button, because a credential anyone can buy is not a credential.
Applications close 11 September.