Sub-function
Automation Factory
Owns automation work end to end inside AI GBS & Global Business Services. Runs 4 lines across 26 stations, drains the repeatable share into the AI-OPS tower, and holds 4 human gates.
Estate rating
A
75 / 100
Build and test are agentic. Putting new automation into a controlled process requires a control review — an unreviewed agent in production is how estates rot.
Agents
11
Lines
4
Units / wk
8,788
Timetable vs actual
23.5h → 19.8h
Cost / unit
$11
was $26
Run-rate saving
$7.0M
annualized at this volume
The 11 agents running this sub-function
Grouped by what they are for, not by where they sit. The agent org is process-shaped.
Automation Factory Orchestrator
Owns the automation factory lane end to end. Sequences the other agents, holds the timetable, and decides what surfaces to a human.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Opportunity Mining Agent
Watches automation factory continuously. Detects drift, quantifies it, and routes what matters without waiting for a reporting cycle.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Regression Monitor Agent
Watches automation factory continuously. Detects drift, quantifies it, and routes what matters without waiting for a reporting cycle.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Feasibility Scoring Agent
Executes the automation factory step it owns, posts its evidence, and hands the unit to the next stage.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Build Agent
Executes the automation factory step it owns, posts its evidence, and hands the unit to the next stage.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Deployment Agent
Executes the automation factory step it owns, posts its evidence, and hands the unit to the next stage.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Test Harness Agent
Runs the repeatable portion of automation factory inside the shared transaction spine at spine cost and spine controls.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Benefit Realization Agent
Runs the repeatable portion of automation factory inside the shared transaction spine at spine cost and spine controls.
ceiling A3 · escalates to Escalates to Head of Automation when the unit falls outside the decision envelope or confidence drops below the floor.
Automation Portfolio Agent
Authors positions on automation factory. Frames the question, models the options, and writes a recommendation a human can ratify or reject.
ceiling A3 · escalates to Escalates when two options sit inside the confidence band and the evidence cannot separate them.
Build Policy Agent
Holds the rules that bind every other agent in automation factory. Versions them, tests them against live decisions, and blocks work that breaches them.
ceiling A3 · escalates to Escalates on any decision that would breach a live policy version, with the clause cited.
Automation Factory Challenger Agent
Adversarial reviewer for automation factory. Argues the opposite case on every position and flags where the evidence does not carry the claim.
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.
People
What the humans stopped doing, and what they do now.
Redeployed into data stewardship and evidence quality.
Lines running in this sub-function
Each line is a workflow. Click a line to walk it station by station.
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.
Automation Portfolio Position
Ratifiedv1Which parts of automation factory should deliberately stay human, and how do we defend that against a cost argument?
Hold the ceiling at A3 for automation factory. Build and test are agentic. Putting new automation into a controlled process requires a control review — an unreviewed agent in production is how estates rot. Raise it only after two consecutive quarters where the override rate stays under five percent and every override has a written cause.
confidence
65%
The challenge — Automation Factory Challenger Agent
Automation Factory 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 10 percent for two consecutive months
- ·Cost per automation 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 $38k are sampled, not fully measured
| Alternative considered | Cost | Risk | Verdict |
|---|---|---|---|
| Hold the current position | $659k run rate | Known and priced. Cedes ground if comparators move faster. | not selected |
| Raise the ceiling one level now | $403k to build controls | Override rate is 13 percent; raising the ceiling before that settles imports the error into production. | rejected on evidence |
| Move the exception tail to the spine | $632k transition | Loses local context. Rework risk on the cases that are hardest to recover. | selected |
Ratified 6 months ago. Measured since: cost per automation fell 7 percent against a forecast of 13 percent.
Automation Factory Operating Position
Draftv2Where should the autonomy ceiling for automation factory sit over the next four quarters, and what evidence would justify moving it?
Keep the repeatable automation 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
83%
The challenge — Automation Factory Challenger Agent
Automation Factory Challenger Agent argues the recommendation leans on 2 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 10 percent for two consecutive months
- ·Cost per automation 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 considered | Cost | Risk | Verdict |
|---|---|---|---|
| Hold the current position | $843k run rate | Known and priced. Cedes ground if comparators move faster. | live |
| Raise the ceiling one level now | $476k to build controls | Override rate is 14 percent; raising the ceiling before that settles imports the error into production. | rejected on evidence |
| Move the exception tail to the spine | $666k transition | Loses local context. Rework risk on the cases that are hardest to recover. | under review |
Policy register
The rules the agents above are bound by. Version, owner, approver, and the agents each rule constrains.
| Ref | Rule | Scope | Owner agent | Approved by | Version | Status |
|---|---|---|---|---|---|---|
| GB7-POL-200 | Automation Factory Decision Envelope Agents in this lane may act without a human when the automation 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 5 agents · LATAM · US | envelope | Build Policy Agent | Head of Automation | v3.4 | under review |
| GB7-POL-201 | Automation Factory 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 5 agents · EU · UK · US | evidence | Build Policy Agent | Head of Automation | v2.2 | Active |