Reference models · Consumer Goods

What governed autonomy looks like — modeled before it is sold.

Three composite reference models show where governed decision loops matter before anyone is asked to buy a product. Each model names the operating context, the loop to govern, the primitives required, and the reason an evaluator should care.

Multi-plant packaged food network
01
Operating domain
Plan / Make / Deliver
Target audience
Supply planning, plant operations, and planning transformation teams
PDF reference
Section 1 · Replenishment governance

Replenishment governance · multi-plant packaged food

A stock-imbalance model for turning shortage and excess signals into governed replenishment actions across plants.

Use this if...

you need to coordinate replenishment actions across plants without hiding policy, exception rights, or escalation logic inside planning workarounds.

Decision loop preview

Sense inventory, service, capacity, and timing signals; decide the prepared action class; optimize within constraints; govern approval tier and autonomy level before execution.

Governance angle

Shows how earned autonomy can be limited by decision class, materiality, and exception history instead of being granted as a blanket automation permission.

Decision loops

  • Sense SKU-site shortages, excess stock, service pressure, and production constraints.
  • Decide whether to rebalance, expedite, defer, substitute, or escalate.
  • Execute only with decision-right evidence and a retained override trail.

Governance primitives

  • Decision rights
  • Approval tiers
  • Autonomy thresholds
  • Override trail
Launch and promotion allocation under constrained supply
02
Operating domain
Plan / Deliver / Finance
Target audience
Commercial supply, customer service, revenue management, and S&OP leaders
PDF reference
Section 2 · Contested service commitments

Allocation under contested service commitments · beauty & personal care

A contested-allocation model for making launch, promo, and scarce-stock tradeoffs explicit before planners are forced into manual arbitration.

Use this if...

scarce launch or promo inventory creates repeated tradeoffs between service promises, strategic customers, channel priorities, and margin exposure.

Decision loop preview

Sense constrained supply and competing commitments; decide the allocation policy; optimize service and financial tradeoffs; govern the exception path before orders are released.

Governance angle

Moves customer-tier logic, channel priority, and margin tradeoffs into inspectable policy so overrides are deliberate and reviewable.

Decision loops

  • Sense demand spikes, available-to-promise gaps, margin exposure, and customer-tier commitments.
  • Decide allocation policy across channels, accounts, launch waves, and service promises.
  • Learn from overrides, substitutions, and service exceptions after each governed run.

Governance primitives

  • Policy constraints
  • Scenario evidence
  • Human approval envelope
  • Audit-ready rationale
Reusable operating-layer pattern
03
Operating domain
Govern / Finance / Enterprise orchestration
Target audience
Technical evaluators, governance reviewers, and operating-model owners
PDF reference
Section 3 · Governed-autonomy reference pattern

The governed-autonomy reference pattern

A reference pattern for evaluating whether an agentic operating layer has the primitives required to coordinate governed decisions safely.

Use this if...

you are comparing agentic planning, orchestration, or optimization claims and need a practical checklist for governed autonomy.

Decision loop preview

Sense the decision class; decide the allowed autonomy envelope; optimize with traceable assumptions; govern exception handling; learn only from retained evidence.

Governance angle

Defines the non-negotiable primitives a governed operating layer must expose before higher autonomy can be considered credible.

Decision loops

  • Classify recurring decision archetypes before assigning automation rights.
  • Attach governance primitives to the loop, not only to the model or workflow.
  • Use execution traces and override history to adjust future autonomy levels.

Governance primitives

  • Identity and context
  • Decision envelope
  • Simulation record
  • Rollback and override semantics
Continue the trace

One combined PDF contains all three models; card-level references point to the relevant section.