Last week I wrote about a simple way to decide what HR should actually build with AI: compare ideas by impact and effort before jumping into tools, vendors or pilots.
A lot of people replied with questions about the step before that. One of my favorite questions was basically this: once you have dozens of plausible AI use cases, how do you cut them down to the few worth serious investigation?
I used ChatGPT to build the HR AI Opportunity Map to answer that exact question. It starts with more than 70 organizational HR AI use cases and lets you narrow them by HR function, potential value, implementation effort and the level of access each build normally needs.
Click here to narrow 70+ HR AI use cases to your best few — free →Likely top use cases
Cut the list fast
Function is the easiest cut. If you are focused on Talent Acquisition, remove everything else.
Then compare what is left by potential value and implementation effort. That quickly separates small, easy builds from projects that may create much more value but need more data, integration and organizational work.
Then look at what the build actually needs. Some can run on content you already have. Others need structured HR data or live access to systems.
An interview guide builder may only need a job description, competencies and interview standards. Interview scheduling needs recruiting data plus live access to recruiting and calendar systems.
The idea may sound equally simple in a list of “AI use cases.” The build is not.
Then inspect the strongest few
Once you have only a handful left, you should go deep on each.
For each use case, the tool shows what it does, its potential value, implementation effort, access required, how AI is being used, how you can check the output, and current adoption.
Take an HR policy assistant. It works from approved policies, and the answer can be checked directly against the source document.
Now compare that with attrition prediction. It needs historical workforce data, modelling and governance. You also cannot verify a prediction immediately. You have to wait and compare it with what actually happens.
I made the judgment calls in the tool visible rather than pretending they are exact science. Value and effort are directional assessments. Access shows the deepest level normally required. Adoption uses SHRM data where there is a close match and directional estimates elsewhere.1
The point is to make the early decisions quickly, so you can spend real time on the three, four or five ideas that deserve it.
Click here to use the free HR AI Opportunity Map yourself →
Best,
Henry Kaerki
Founder, Wellhana