Skip to content
Empley

HR agents

One agent, one job, done well.

Ten HR agents run inside your environment, on your data. Each one states its trigger, inputs, outputs and who approves what.

  • Core HR process
  • Planning and cost
  • Documents and risk
  • Insight and analytics

10 agents · 8 packages · 3 models

A curved metal mesh catching warm light.
Core HR process
Planning
Documents
Insight
Heavy wooden doors with brass ring handles.

System map

Four domains, one governed system.

Core process, planning, documents and insight stay visually separate, so roster, sketches and guardrails are easy to scan.

Not an agent gallery. A governed system with distinct roles.

Roster

Ten agents. Each with one clear job.

Onboarding Agent

Standalone

Runs the first 30 days as one tracked sequence, with owners per task.

Lives inAgent Operations

Screening Agent

Standalone

Structures every application against the same agreed criteria.

Lives inAgent Operations

Recruiter decides, always

HR Service Desk Agent

Standalone

Answers policy questions from your own documents, escalates the rest.

Lives inAgent Operations

Skills Profiling Agent

In package

Turns roles, tasks and evidence into one confirmed skills picture.

Lives inSkills and Task Model

Needs the skills and task model

Attrition Signal Agent

In package

Shows where leaving rates run above trend, and which driver explains it.

Lives inPeople Analytics and Metric Layer

Needs the metric layer

Workforce Forecast Agent

In package

Keeps demand and internal supply on the same monthly clock.

Lives inDemand and Supply Forecasting

Lives in the forecasting module

Labour Cost Agent

Standalone

Answers workforce cost scenarios with the assumptions shown.

Lives inDemand and Supply Forecasting

Adoption Agent

Standalone

Measures real use of new ways of working, team by team.

Lives inImplementation and Fast-to-Impact

Policy and Compliance Agent

Standalone

Checks documents and cases against the rule in force, with citations.

Lives inReadiness and Risk Control

Job Architecture Agent

In package

Turns titles and tasks into role families and written level criteria.

Lives inSkills and Task Model

Runs inside the job architecture work

Every package is built from this roster. Want just the agent? Several run standalone. See the packages.

Separate roles, one flow.
A calm room with soft daylight and few objects.
Every run leaves a readable trace
Demo

See an agent run before you build.

Clickable sketches with inputs, steps, outputs and guardrails. Illustrative only. Real runs go against your systems.

How an agent is built

Four steps. No mystery.

  1. 01

    Diagnose

    We map the process, the systems and the data. Two weeks. You get a map.

  2. 02

    Sketch

    A deterministic sketch you can click. Trigger, inputs, steps, outputs, guardrails.

  3. 03

    Pilot

    Built against your systems, run on real data, inside your security model.

  4. 04

    Operate

    The agent runs in your environment. You own it. We stay for iteration.

  • Runs in your environment
  • EU data residency
  • Audit trail per step
  • Guardrails approved by you
Moss set inside hexagonal wall frames.
Every step gets a status you can follow afterwards.
Dark green leaves in low light.

Under the surface

Every step can be read back.

Input, rule, decision and outcome are logged per run. You see what the agent did, when, and on what evidence.

The controls are yours, visible from day one.

Try it

ILLUSTRATIVE

What would your agent do?

Describe the bottleneck. You get an indication of which agent fits, and the next step. No invented numbers.

Talk to a person
Next step

Book half an hour. You will know what is sensible to start with.

Get in touch

Say what you want done, we come back with a plan.

Compare the packages

Three entry points, five modules, the same terms underneath.

Read about security and ownership

Full log per run, a stop switch, and the right to take everything with you.