Twenty-five years in the People team space, now teaching those teams to get more done with AI
I am Mark North. Twenty-five years in hiring, leading talent acquisition and working shoulder to shoulder with People teams, mostly in the kind of companies this site is written for. I am not a technologist who discovered HR, and that matters: the quickest way to lose a room of People professionals is to explain AI the way an engineer would.
What this is
People Team AI teaches People teams to understand AI well enough to diagnose a problem, test the simplest sensible change and say whether it worked. The training stands on its own and most people buy only that.
Where an organisation wants a wider operating method around that capability, the PROVEN Framework™ is there, and TransformLayer is the software the programme runs in. People Team AI trains the people. The PROVEN Framework™ installs the method. TransformLayer runs the programme. Each is bought on its own.
Anyone can learn AI. The difficulty is the volume of it, and the analysis paralysis that follows. Whether your team uses ChatGPT, Claude, Gemini, Copilot or something else, the training is built around what you already have, and it turns People professionals into the AI champion the function needs.
Training runs in person, or online for remote and hybrid teams. If you want to work out which suits yours, email me and we can go through the options.
923 occupations, 19,265 tasks, every source named
The exposure report says of each task whether a chat window could already do it, whether it would need software building first, or whether it cannot be sped up at all. 771 occupations have a page of their own, covering roughly 159 million people in work in the United States.
None of the underlying ratings are mine. They come from published research, they are named on every page, and the versions are stated so anybody can reproduce or contradict the numbers.
The method and its limits, in full, including where it should not be trusted.
The caveat that matters. This is task-level exposure, not a prediction about anybody's employment. A task a model can reach is not a job it can do, and a language model cannot do physical work, which is most of why manual trades score the way they do. It is not a ranking of which careers are worth having.
Getting hold of me
Questions, corrections and challenges to the data reach me at email me . If something here is wrong I would genuinely rather know, and corrections get published rather than quietly fixed.
For press, podcasts and speaking, the same address reaches me. I am happy to talk through the method, the limits, or any single occupation in the report.
Elsewhere
Mark North on LinkedIn, which is where the findings get posted as they come out.
The research, the methods, the benchmark material and the free tools are published openly and free to take. Training, consulting and TransformLayer are commercial services built on that work. Nothing on this site is sponsored, and no vendor pays to appear in it.
If you got here because you are weighing up whether to bring someone in to teach your team, that is what the workshops are. Everything on this page is true whether you book one or not.