OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
BlueHourMOM-304 · COMPLEXITY CEILING MANAGEMENT

More intelligence. How much of it can your enterprise put to work?

Follow Larkspur Industrial Supply, a fictional distributor, as it adds AI agents to order to cash. Demand stays fixed at 6,000 orders a week. You change three things: how many decisions the agents make, how tightly the functions depend on each other, and the operating architecture.

MOM-304 is in design. Every number here is illustrative and comes from a simple model you can inspect.

YOU CHANGEDecisions a dayHow much work the agents take on
YOU CHANGECouplingHow tightly functions depend on each other
YOU CHANGEArchitectureFive design choices in the operating model
STAYS FIXEDDemand6,000 orders a week, so before and after compare fairly
Designed with Physics. Deployed with Economics. Determined by Humans.
BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 1 · THE ENTERPRISE MAP

Who intends what, what depends on what, and who owns it

Larkspur Industrial Supply, a fictional distributor. Order to cash, with demand fixed at 6,000 orders a week.
Larkspur order-to-cash dependency mapNine nodes with owners and intents. Sales reserves inventory, asks credit to check the customer and pricing for a discount; pricing quotes customers. Inventory and credit both feed fulfillment, which invoices through finance. Two feedback loops: inventory and procurement, finance and credit. Fulfillment and finance escalate to the COO.reservereorderdiscountquotecredit checkpickreleaseinvoicepayment historyescalateSales agentsOwner: VP SalesIntent: Grow ordersInventoryOwner: Supply chain directorIntent: Keep stock availableProcurementOwner: CPOIntent: Buy at lowest costPricingOwner: Pricing leadIntent: Protect marginCreditOwner: Credit managerIntent: Limit bad debtFulfillmentOwner: VP OperationsIntent: Ship on timeCustomersOwner: Commitments madeIntent: On-time deliveryFinanceOwner: ControllerIntent: Collect cashAccountable executiveOwner: COOIntent: Decides escalations
Dependency, one directionFeedback loopEscalation to the accountable executive

Not every function depends on every other. What matters is the shape: which decisions wait on which, where feedback loops can amplify a mistake, and who owns each intent. Today the agents make about 2,000 decisions a day and the process completes 2,910 of the 6,000 orders demanded right the first time.

BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 2 · ADD INTELLIGENCE

Useful work grows, and so does coordination demand

This step starts at 5,000 agent decisions a day, up from 2,000. Move the slider to see both effects.
WHAT THE AGENTS DO, AND WHAT THE ENTERPRISE COMPLETES
Agent decisions this week0
Every decision an agent makes across sales, pricing, inventory, procurement and credit.
Orders the process can handle0
Capacity to process orders. The marker shows the 6,000 demanded.
Useful completed work0
Orders done right the first time. This is the number that reaches the income statement.
COORDINATION DEMAND AGAINST CAPACITY · CROSS-FUNCTION DECISIONS A DAY
Coordination demand0
Decisions that touch another function’s commitment: stock promised, credit used, a truck booked.
Coordination capacity0
What the current operating model can resolve without creating an exception.
THE ECONOMIC CEILING COMES FIRST

Activity is not output. Useful work rises at first, because agents take work people could not reach. Coordination demand rises with it, and the cost of each extra useful order rises long before anything fails.

BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 3 · INTERFERENCE

The same accelerated order, two ways

Individually successful decisions can reinforce each other or work against each other. The difference is architecture, not effort.
REINFORCING · CONSTRUCTIVE
  1. SalesAn agent accepts an accelerated order for 400 units, due Friday.
  2. InventoryConfirms the stock is unreserved, then reserves it.
  3. CreditHas already cleared the customer’s limit. The order proceeds.
  4. FulfillmentBooks Friday’s truck with the commitment visible.
  5. ProcurementSees a planned draw-down, not a shortage. Buys nothing extra.
One order, one commitment, shipped on time. Every decision added to the result.
CONFLICTING · DESTRUCTIVE
  1. SalesAn agent accepts the same order using stock already reserved for another customer.
  2. ProcurementSees inventory go negative and places a rush buy at a premium.
  3. CreditHolds the order: the new total exceeds the customer’s limit.
  4. FulfillmentOpens two exceptions: one held order, one broken promise.
  5. FinanceWaits on cash from both customers while carrying the rush stock.
Every agent did its job. Activity rose, and enterprise output fell.
THIS WEEK, AT THE SETTINGS IN THE PANEL

0 cross-function decisions fitted together. 0 became exceptions, from conflicts amplified by feedback or decisions beyond capacity. Procurement’s reactions bought an estimated $0 of unplanned inventory, working capital tied up rather than a loss.

Constructive and destructive interference are an analogy from physics: aligned contributions add up, opposed ones cancel. Enterprise decisions do not obey a wave equation. The analogy is about alignment of intent, incentives, timing and authority.

BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 4 · APPROACHING THE LIMIT

The warning signs come before the outage

8,000 agent decisions a day, the same architecture, the same demand. Four times the decisions of step 1.
QUEUES
0
Exceptions a week, orders that need a person
REWORK
0
Hours of human exception work a week
DECISION DELAY
0
Median wait for an exception decision
RECOVERY MARGIN
0
Time to recover from a 2-hour credit-system outage
ACTION THRESHOLDS · SET BY THE ACCOUNTABLE EXECUTIVE BEFORE THE LIMIT IS NEAR
Margin above 25% keep expanding at the planned pace15 to 25% slow expansion and start redesignBelow 15% the COO decides: pause, redesign or reduce loadBelow zero, or no recovery controlled intervention, MOM-305

These thresholds are illustrative. In practice they are set per workflow, from what that process can absorb, and written down before anyone needs them.

Nothing has failed yet. Every agent is working. The warning signs are queues, rework, slower decisions and a recovery that no longer completes. They arrive well before an outage, which is the point of watching them.

BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 5 · REDESIGN AND RETEST

Change the architecture, then retest at the same demand

Still 8,000 decisions a day. Turn the five changes on one at a time and watch the panel.
CHANGE TO THE OPERATING MODELEFFECT IN THIS MODELRUN COST
THE TRADEOFF

All five cost $27K a week to run and hold $1.2M of working capital in reserved inventory and credit. Designing them takes owner time this demonstration does not price. Only one of the five constrains the agents, and it raises capacity rather than lowering volume. None of them is an approval gate.

BlueHourENTERPRISE OPERATING MODEL
How the numbers workLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
STEP 6 · THE EXECUTIVE DECISION

Expand, redesign first, or intervene

Before and after the architecture changes, at the same demand. Then test the envelope under degraded conditions.
SAME DEMAND · 8,000 AGENT DECISIONS A DAY · MODERATE COUPLING · PER WEEK, ILLUSTRATIVE
Current architectureRedesigned
Useful completed work2,940 orders5,699 orders
Exception workload3,060 · 2,295 h301 · 120 h
Operating cost$312K · $106 per useful order$58K · $10 per useful order
Revenue at risk (estimate)$5.5M$45K
Time to decide50 h and growing4.0 h
Time to recoverDoes not recover without intervention5.3 h
Coordination margin-33%38%
Architecture run cost$0$27K + $1.2M buffer held
SATURATION TEST · THE OPERATING ENVELOPE
Test condition
Current architectureRedesignedPanel settings

Boundary: where coordination margin falls to 15%, an illustrative threshold. The COO authorizes expansion on evidence from this test: at least 25% margin under the degraded condition, recovery inside tolerance. MOM-304 shows how close you are to the edge; MOM-305 determines what happens when you reach it →

OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
OPERATING LEVERAGE

Put more intelligence to work, inside limits you have tested.

The goal is more useful output from the resources committed, with capital, customers and continuity protected. MOM-304 is in design, and it is activated after MOM-001, Capital Discipline.

More intelligence, more coordinationAgents can take on far more work. Every decision that touches another function adds to the coordination the enterprise must do.
Architecture decides the outcomeThe same agents, the same demand: 2,940 useful orders a week or 5,699, depending on how decisions fit together.
Limits need evidence and an ownerTest the operating envelope before you expand it, and name who may authorize expansion and on what evidence.
MOM-304 shows how close you are to the edge.MOM-305 determines what happens when you reach it.

The agent is not the system. More intelligence requires more architecture, not less.

PUT MOM-304 ON YOUR ROADMAP

Find out how much more intelligence your enterprise can put to work, and what architecture it needs first.

Request MOM-304 →
BlueHour
Designed with Physics. Deployed with Economics. Determined by Humans.
Opus. Humanitas. Veritas. Work. Humanity. Truth.
START WITH MOM-001bluehourbusiness.com WHEN ACTION IS NEEDEDbluehourrisk.com WORKING WITH BLUEHOURbluehourtech.com Return to the beginning
BlueHourENTERPRISE OPERATING MODEL
Back to the startLarkspur IndustrialRD
OPERATING MODEL AS A SERVICE · DEMONSTRATION · MOM-304 IN DESIGN · ILLUSTRATIVE DATA, FICTIONAL COMPANY
REFERENCE · THE DEMONSTRATION MODEL

How the numbers work

A deliberately simple model of a fictional order-to-cash process, written down before it was coded. Demand is fixed at 6,000 orders a week, $1,800 each. All figures are per week. It is not a validated or proprietary complexity score.
A1
Orders the process can handle2,400 + 0.45 × agent decisions a day, up to the 6,000 demanded.
A2
Coordination demandAgent decisions × share that touch another function (15%, 30% or 50% by coupling), × 0.7 if dependent decisions are sequenced.
A3
Coordination capacity1,800 decisions a day, raised 25% by reserved capacity, 15% by boundaries, 10% by clear authority; reduced under a degraded condition.
A4
Coordination margin1 − demand ÷ capacity. A demonstration index, not a measured score.
A5
Conflict rate10% of coordinated decisions; × 0.4 with shared objectives; × 0.8 with boundaries; × 2.5 if conflicting targets are reintroduced.
A6
Feedback amplification× 1.0, 1.3 or 1.8 by coupling; the excess is cut 60% by boundaries.
A7
ExceptionsAmplified conflicts, plus half of any decisions beyond capacity, over five days.
A8
Useful completed workOrders the process handles, minus exceptions.
A9
Exception workload0.75 hours each, or 0.4 with clear authority.
A10
Time to decide4 hours (1.5 with clear authority) ÷ (1 − load), a simple queueing approximation. At full load the backlog grows daily.
A11
Revenue at riskEstimate: exceptions likely to miss the customer commitment × $1,800. Not a realized loss.
A12
Unplanned inventoryEstimate: purchases triggered by amplified conflicts, $600 each. Working capital, not a loss.
A13
Time to recoverFrom a 2-hour credit-system outage: 2 ÷ (1 − load) hours. At full load it does not recover without intervention.
A14
Operating cost$0.25 per agent decision, $85 per exception hour, $35 rework per exception, plus architecture run cost.
A15
Reserved buffer$1.2M of working capital held when reserved capacity is on. Shown separately, never added to weekly cost.
WHICH INPUTS MOVE WHICH OUTPUTS
Decisions a dayCouplingArchitectureDegraded test
Useful completed work●●●●
Exception workload●●●●
Operating cost●●●●
Revenue at risk●●●●
Time to decide●●●●
Time to recover●●●●
Coordination margin●●somesome

“Some”: shared objectives and a reintroduced conflict change exceptions and output, not coordination margin.

Reset in any step returns that step to its starting values. The figures always come from the same formulas, so the same settings give the same numbers.

Go to step 1