Skills Supply & Demand Agent
A skills supply & demand agent is an AI agent for workforce planning & org design that builds a current skills inventory from job architecture, learning records and opt-in employee profiles, compares it with the skills the strategy and hiring plan will require, and shows where the gaps and surpluses sit.
How does the skills supply & demand agent work?
What flows in, what the agent does with it, where a person decides, and what comes out.
Reads from
Job architecture and skills taxonomy · Learning and certification records · Opt-in employee skill profiles · Hiring plan and strategy priorities · Employee opt-in and visibility rules
AI agent · runs when the workforce-plan refresh, or a leader asks
Skills Supply & Demand Agent
A person decides
Talent and L&D leaders decide build, buy or borrow per gap
Produces
Skills gap and surplus heatmap · Build, buy or borrow options per gap · Where each inference came from
What does the skills supply & demand agent do?
Builds a current skills inventory from job architecture, learning records and opt-in employee profiles, compares it with the skills the strategy and hiring plan will require, and shows where the gaps and surpluses sit.
What does it produce?
A skills gap heatmap by function and location with build, buy or borrow options for each gap
Who decides?
Talent and L&D leaders decide whether to build, buy or borrow for each gap; the agent infers and compares, and any use of individual profiles is opt-in and visible to the employee.
What systems does the skills supply & demand agent connect to?
Examples of the kind of systems this agent would read from or write to, so you can picture it in your own stack. The actual set is whatever you run.
-
HRIS / skills
job architecture and taxonomy
-
Learning
learning records
-
Talent marketplace
opt-in profiles and interests
-
ATS
planned roles and criteria
What data does it need?
- job architecture and skills taxonomy
- learning records
- opt-in employee skill profiles
- hiring plan and strategy documents
How would you measure it?
opt-in coverage by function, monthly; time-to-close each named critical gap, per gap; share of open roles filled internally, quarterly
What does a first proof look like?
Pick one function with a volunteer group who opt in. In a sandbox the agent builds the inventory from taxonomy, learning records and those profiles, then compares it with the next two quarters' hiring plan. Function leaders and L&D check the heatmap against what they know.
You'd call it working when
The gaps match the ones they already feel, and every employee can see and correct what was inferred about them.
What usually goes wrong?
- An unmaintained taxonomy produces confident nonsense; refresh it before the agent reads it
- Skills inferred from course completions overstate proficiency; show the source of every inference
- Opt-in collapses if employees suspect profiles feed performance or redundancy decisions
What are the guardrails?
- Individual profiles are opt-in, visible and correctable by the employee; nothing inferred is hidden from them
- Outputs at function and location level only, with a minimum group size below which cells are suppressed
- Profiles never feed performance, pay or redundancy processes
- GDPR profiling notice, DPIA and DSAR-ready export; works-council or employee-representative consultation before launch
- Skills inferred from learning records are labeled as inferred, with the source
What leaves your boundary is set per build; the inputs above are the ceiling, and where the model runs, what it retains, and the DPA are agreed with your security team before anything is connected.
Our read
Plausible and interesting, few real deployments yet. Worth a proof of concept before a production budget.
Little of this exists off the shelf yet, which is part of why it is a proof before it is a build.
inferring skills for named individuals is profiling under GDPR and a works-council matter
Where it sits in the order
Depends on a maintained taxonomy, employee trust and consultation; the data is rarely ready and the profiling risk is real
Is a Skills Supply & Demand Agent worth building for your function?
That depends on your numbers, your data, and what else is on the map for you. The strategy month works that out.
Thirty minutes. Bring the number this would move.