LensReading which lens this session carries.
Customer Service/Knowledge Management

Sub-function

Knowledge Management

Owns article work end to end inside Customer Service. Runs 3 lines across 17 stations, drains the repeatable share into the AI-OPS tower, and holds 3 human gates.

orchestrator Knowledge Orchestratorhuman owner Head of Knowledge Management routes into ai-ops

Estate rating

A

80 / 100

autonomy 71%cap 80%

Drafting and localization are agentic. Publishing an answer customers will act on requires a human accuracy sign-off — wrong knowledge scales instantly.

Agents

10

Lines

3

Units / wk

8,585

Timetable vs actual

8.5h → 8.9h

Cost / unit

$13

was $43

Run-rate saving

$13.3M

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.

Knowledge Orchestrator

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

ObserveA3185 / 24h94%9% ovr$12/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

ObserveWatches signals and watches the agents.

Gap Detection Agent

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

ObserveA317 / 24h89%10% ovr$13/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

Usage Analytics Agent

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

ObserveA321 / 24h90%8% ovr$7.35/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

TaskExecutes stations on a line.

Article Authoring Agent

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

WorkflowA33,732 / 24h97%9% ovr$0.60/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

Accuracy Verification Agent

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

WorkflowA3547 / 24h88%9% ovr$0.16/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

Localization Agent

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

WorkflowA33,804 / 24h92%9% ovr$0.52/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

AI GBSRuns the shared transaction spine.

Freshness Agent

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

AI GBSA31,283 / 24h91%6% ovr$0.16/task

ceiling A3 · escalates to Escalates to Head of Knowledge Management when the unit falls outside the decision envelope or confidence drops below the floor.

PolicyAuthors and versions the binding rules.

Knowledge Policy Agent

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

PolicyA256 / 24h90%5% ovr$14/task

ceiling A3 · escalates to Escalates on any decision that would breach a live policy version, with the clause cited.

StrategyForms and defends a position.

Knowledge Doctrine Agent

Authors positions on knowledge management. Frames the question, models the options, and writes a recommendation a human can ratify or reject.

StrategyA2218 / 24h87%12% ovr$11/task

ceiling A3 · escalates to Escalates when two options sit inside the confidence band and the evidence cannot separate them.

ChallengerPaid to disagree before a human has to.

Knowledge Management Challenger Agent

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

StrategyA2186 / 24h95%16% ovr$10/task

ceiling A3 · 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 Coverage76
Evidence Quality82
Override Discipline84
Data Integrity94
Recovery Readiness86
Cost Transparency90
Judgment Readiness55

People

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

headcount on this work62.739.3 FTE

Redeployed into cross-function process redesign.

inference cost per unit$3.47

Lines running in this sub-function

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

11 clear4 evidenced2 held
CS5AGap to Published ArticleTransactional · 6 stations

Detect gap

4 live

Draft article

1 live

Verify accuracy

1 live

Localize

20 q

Publish

2 live

Measure

6 q

CS5BFreshness CycleTransactional · 5 stations

Scan staleness

3 live

Re-verify

1 live

Update draft

17 q

Approve

2 live

Republish

1 live

CS5CKnowledge Doctrine ReviewJudgment · 6 stations

Frame question

31 q

Gather evidence

1 live

Model options

12 q

Draft position

1 live

Challenge

1 q

Ratify

0 q

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.

CS5-POS-100

Knowledge Doctrine Review

In challengev1

Where should the autonomy ceiling for knowledge management sit over the next four quarters, and what evidence would justify moving it?

Hold the ceiling at A3 for knowledge management. Drafting and localization are agentic. Publishing an answer customers will act on requires a human accuracy sign-off — wrong knowledge scales instantly. Raise it only after two consecutive quarters where the override rate stays under five percent and every override has a written cause.

confidence

70%

The challenge — Knowledge Management Challenger Agent

Knowledge Management Challenger Agent argues the recommendation leans on 5 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 article stops falling while volume keeps rising
  • ·A control failure in this lane reaches a customer or a regulator

Evidence gaps

  • ·No independent test of the control evidence behind the touchless rate
Alternative consideredCostRiskVerdict
Hold the current position$886k run rateKnown and priced. Cedes ground if comparators move faster.live
Raise the ceiling one level now$708k to build controlsOverride rate is 8 percent; raising the ceiling before that settles imports the error into production.rejected on evidence
Move the exception tail to the spine$1042k transitionLoses local context. Rework risk on the cases that are hardest to recover.under review
authored by Knowledge Doctrine Agentratifier Head of Knowledge Management3 stated assumptions
CS5-POS-101

Knowledge Management Operating Position

Supersededv3

Should article work stay in the shared transaction spine, or come back inside the function where the context lives?

Keep the repeatable article volume in the spine and pull the exception tail back into the function. The spine is cheaper per unit; the tail is where the context that prevents rework actually sits.

confidence

88%

The challenge — Knowledge Management Challenger Agent

Knowledge 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 8 percent for two consecutive months
  • ·Cost per article stops falling while volume keeps rising
  • ·A comparator publishes a materially lower unit cost on a like-for-like basis

Evidence gaps

  • ·No independent test of the control evidence behind the touchless rate
Alternative consideredCostRiskVerdict
Hold the current position$74k run rateKnown and priced. Cedes ground if comparators move faster.live
Raise the ceiling one level now$427k to build controlsOverride rate is 7 percent; raising the ceiling before that settles imports the error into production.rejected on evidence
Move the exception tail to the spine$350k transitionLoses local context. Rework risk on the cases that are hardest to recover.under review
authored by Knowledge Doctrine Agentratifier Head of Knowledge Management3 stated assumptions

Policy register

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

RefRuleScopeOwner agentApproved byVersionStatus
CS5-POL-200

Knowledge Management Decision Envelope

Agents in this lane may act without a human when the article 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 · APAC

envelopeKnowledge Policy AgentHead of Knowledge Managementv3.3Active
CS5-POL-201

Knowledge 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 · EU · UK

evidenceKnowledge Policy AgentHead of Knowledge Managementv4.2Active