LensReading which lens this session carries.
Human Resources/Performance Management

Sub-function

Performance Management

Owns review work end to end inside Human Resources. Runs 3 lines across 19 stations, drains the repeatable share into the H2R tower, and holds 6 human gates.

orchestrator Performance Orchestratorhuman owner Director, Performance & Talent routes into h2r

Estate rating

BBB

70 / 100

autonomy 49%cap 55%

Agents assemble evidence, scan for biased language and prepare calibration packs. Ratings, PIP initiation and outcomes are never agent decisions — they carry legal consequence.

Agents

10

Lines

3

Units / wk

13,120

Timetable vs actual

15.4h → 16.4h

Cost / unit

$17

was $33

Run-rate saving

$11.1M

annualized at this volume

The 10 agents running this sub-function

Grouped by what they are for, not by where they sit. The agent org is process-shaped.

OrchestratorOwns the queue and arbitrates between desks.

Performance Orchestrator

Owns the performance management lane end to end. Sequences the other agents, holds the timetable, and decides what surfaces to a human.

ObserveA1107 / 24h96%8% ovr$11/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

AI GBSRuns the shared transaction spine.

Goal Cascade Agent

Runs the repeatable portion of performance management inside the shared transaction spine at spine cost and spine controls.

AI GBSA12,055 / 24h96%6% ovr$0.21/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

TaskExecutes stations on a line.

Continuous Feedback Agent

Executes the performance management step it owns, posts its evidence, and hands the unit to the next stage.

WorkflowA12,593 / 24h91%2% ovr$0.28/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

Calibration Prep Agent

Executes the performance management step it owns, posts its evidence, and hands the unit to the next stage.

WorkflowA13,955 / 24h93%7% ovr$0.41/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

PIP Drafting Agent

Executes the performance management step it owns, posts its evidence, and hands the unit to the next stage.

WorkflowA13,695 / 24h97%5% ovr$0.47/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

ObserveWatches signals and watches the agents.

Review Quality Agent

Watches performance management continuously. Detects drift, quantifies it, and routes what matters without waiting for a reporting cycle.

ObserveA1188 / 24h87%16% ovr$8.90/task

ceiling A1 · escalates to Escalates to Director, Performance & Talent when the unit falls outside the decision envelope or confidence drops below the floor.

PolicyAuthors and versions the binding rules.

Rating Distribution Monitor

Holds the rules that bind every other agent in performance management. Versions them, tests them against live decisions, and blocks work that breaches them.

PolicyA186 / 24h88%11% ovr$7.05/task

ceiling A1 · escalates to Escalates on any decision that would breach a live policy version, with the clause cited.

Performance Policy Agent

Holds the rules that bind every other agent in performance management. Versions them, tests them against live decisions, and blocks work that breaches them.

PolicyA1111 / 24h89%13% ovr$11/task

ceiling A1 · escalates to Escalates on any decision that would breach a live policy version, with the clause cited.

Bias Language Monitor

Holds the rules that bind every other agent in performance management. Versions them, tests them against live decisions, and blocks work that breaches them.

PolicyA1207 / 24h93%4% ovr$6.50/task

ceiling A1 · escalates to Escalates on any decision that would breach a live policy version, with the clause cited.

ChallengerPaid to disagree before a human has to.

Performance Management Challenger Agent

Adversarial reviewer for performance management. Argues the opposite case on every position and flags where the evidence does not carry the claim.

StrategyA2131 / 24h94%5% ovr$11/task

ceiling A1 · escalates to Escalates when a ratified position is still running against a break condition it flagged.

Estate rating breakdown

Seven control dimensions. Judgment readiness runs against autonomy on purpose.

Control Coverage56
Evidence Quality64
Override Discipline87
Data Integrity83
Recovery Readiness70
Cost Transparency62
Judgment Readiness67

People

What the humans stopped doing, and what they do now.

headcount on this work42.534.0 FTE

Redeployed into agent supervision, calibration and coaching.

inference cost per unit$2.33

Lines running in this sub-function

Each line is a workflow. Click a line to walk it station by station.

9 clear3 evidenced1 held
HR5AGoal CycleTransactional · 5 stations

Cascade

37 q

Draft goals

1 live

Manager review

3 q

Lock

2 live

Track

1 live

HR5BYear-end & CalibrationTransactional · 7 stations

Assemble evidence

23 q

Draft reviews

12 q

Bias scan

3 live

Calibration pack

41 q

Calibration session

6 q

Rating lock

1 live

Publish

2 q

HR5CPIP LifecycleTransactional · 7 stations

Trigger

45 q

Evidence pack

1 live

Legal check

3 q

Manager decision

8 q

Plan draft

22 q

Monitor

1 live

Outcome

3 live

Strategy positions this sub-function holds

An agent that only executes is a robot. These are the calls it made, the argument against each one, and what would prove it wrong.

HR5-POS-100

Performance Management Operating Position

Ratifiedv2

Is the current cost per review defensible against external comparators once inference spend is counted in full?

Hold the ceiling at A1 for performance management. Agents assemble evidence, scan for biased language and prepare calibration packs. Ratings, PIP initiation and outcomes are never agent decisions — they carry legal consequence. Raise it only after two consecutive quarters where the override rate stays under five percent and every override has a written cause.

confidence

79%

The challenge — Performance Management Challenger Agent

Performance Management Challenger Agent argues the recommendation leans on 4 quarters of data from a period with no volume shock. The confidence band overlaps the alternative, and the position does not say what it would take to be wrong. Recorded as a dissent, not a block.

Break conditions

  • ·Override rate rises above 7 percent for two consecutive months
  • ·Cost per review stops falling while volume keeps rising
  • ·A control failure in this lane reaches a customer or a regulator

Evidence gaps

  • ·No external comparator on a like-for-like unit definition
  • ·Exception cases under $32k are sampled, not fully measured
Alternative consideredCostRiskVerdict
Hold the current position$188k run rateKnown and priced. Cedes ground if comparators move faster.not selected
Raise the ceiling one level now$250k to build controlsOverride rate is 12 percent; raising the ceiling before that settles imports the error into production.rejected on evidence
Move the exception tail to the spine$220k transitionLoses local context. Rework risk on the cases that are hardest to recover.selected
authored by Performance Orchestratorratifier Director, Performance & Talent3 stated assumptions
outcome held

Ratified 9 months ago. Measured since: cost per review fell 21 percent against a forecast of 20 percent.

Policy register

The rules the agents above are bound by. Version, owner, approver, and the agents each rule constrains.

RefRuleScopeOwner agentApproved byVersionStatus
HR5-POL-200

Performance Management Decision Envelope

Agents in this lane may act without a human when the review sits inside the stated value, risk and confidence envelope. Outside it, the unit holds at a gate with a named approver and a running clock.

binds 4 agents · US · Canada

envelopeRating Distribution MonitorDirector, Performance & Talentv5.3Active
HR5-POL-201

Performance Management Evidence Standard

Every autonomous decision writes inputs, the rule version applied, the model and prompt version, the output and a reversal path. An action with no evidence record is treated as a control failure, not a fast decision.

binds 4 agents · LATAM · US

evidenceRating Distribution MonitorDirector, Performance & Talentv4.8Active
HR5-POL-202

Performance Management Escalation Rule

Escalation is mandatory when confidence falls below the floor, when two options sit inside the confidence band, or when a break condition on a ratified position fires. Director, Performance & Talent owns the response clock.

binds 4 agents · APAC

escalationRating Distribution MonitorDirector, Performance & Talentv3.4Active
HR5-POL-203

Performance Management Autonomy Ceiling

The ceiling for this lane is 55 on the platform scale. Agents assemble evidence, scan for biased language and prepare calibration packs. Ratings, PIP initiation and outcomes are never agent decisions — they carry legal consequence. The ceiling is a governance decision, not a technical limit, and only Director, Performance & Talent can propose moving it.

binds 4 agents · EU · UK · US

ceilingRating Distribution MonitorDirector, Performance & Talentv4.3Active