Agents act. Models decide.
Three models sit under everything we run. Each one states what it needs and what it decides.
Three models
What each model needs, and what it settles.
No model runs without the data underneath it. That is why HR data comes first.
Skills and task model
One taxonomy for what work is done and what it takes to do it.
Needs
- Job architecture and task catalogue
- Learning and certification records
- Employee-confirmed skill profiles
Decides
- Which critical skills are covered and which are not
- Build, hire or borrow per role family
- Which tasks are ready for an agent
Rebuilt monthly, corrected continuously by employees.
Employees see and correct their own profile. No performance inference.
Adoption rate model
Measures real use of new ways of working, not licences bought.
Needs
- An agreed definition of completed work
- Team-level usage and process data
- Training records and pulse feedback
Decides
- Where enablement goes next
- Whether a rollout continues, pauses or changes
- Which blockers cost the most
Weekly during rollout, monthly in steady state.
Team level only, minimum group size ten. No individual league tables.
Demand and supply forecasting
Workforce demand and internal supply on the same clock.
Needs
- Headcount, vacancies and movement history
- Business volume or revenue plan
- Attrition, mobility and time-to-hire rates
Decides
- Headcount plan per role and quarter
- When a gap turns into a delivery risk
- Cheapest credible way to close each gap
Monthly cycle, scenarios on demand.
Every assumption visible, editable and logged with author and date.
Tell us what data you have. We will say which model runs today.
Getting in touch commits you to nothing.
What the models need before they can be trusted.
Ten agents, each with trigger, inputs, outputs and trust tier.