← Shop·The AI Use-Case Library

About this library

Most AI use-case lists are aspirational. This one is honest.

Every HR function I have worked with in the past eighteen months has been handed a list of AI use cases by someone: an internal innovation team, a consulting partner, a vendor roadmap. The lists are long. They are confident. They are, in my experience, mostly wrong — not in the obvious way of suggesting things that do not work, but in the subtler way of suggesting things that work technically and make the function measurably worse.

This library contains seventy-four use cases I have either implemented, watched implemented, or deliberately not implemented. Each one gets three honest verdicts: what it does, what it actually costs (beyond the invoice), and whether I would recommend it to someone whose work I respect. The recommendations are not consistent — the same use case I advise against for one organisation, I might endorse for another. Context matters, and the library tells you why.

The library is updated twice a year. When a use case changes — because a vendor improved it, because the regulatory landscape shifted, because I watched one fail in production — the entry gets rewritten. Buyers of earlier versions receive the updates free of charge.

It is not a buyer's guide. It is not a certification. It is a collection of opinions I would be willing to argue for, sharpened by twenty years of watching these programmes unfold. If you disagree with a verdict, the library is doing its job.

What's in the library

Part I — Recruitment & hiring · fourteen use cases, from sourcing to onboarding
Part II — Employee service & operations · eighteen use cases across shared services
Part III — Manager support · eleven use cases aimed at the mid-level
Part IV — Learning & development · nine use cases, plus a warning section
Part V — People analytics · eight use cases, most of which I do not recommend
Part VI — Performance & reward · seven use cases, unusually careful territory
Part VII — Workforce planning · seven use cases, best in class
Appendix — The nine questions I ask before every implementation