09 · Work Allocation

Who does what.

Work Allocation is the mechanism that assigns work to the right actor. It determines whether work should be done by a human, a Twin, an agent, or a system — and ensures the assignment is appropriate for the work's nature and risk.

---

Allocation Model

Work Item
      ↓
Capability Check → Role Check → Capacity Check → Context Check
      ↓
Allocation Decision
      ↓
Actor Assignment
      ↓
Execution

---

Allocation Options

Actor	When Used	Authority Level
Human	Work requiring judgment, creativity, or authority	Full authority
Twin	Work that can be reasoned about and planned	Proposes, human approves
Domain Agent	Work automatable within a business domain	Executes under governance
Specialist Agent	Work requiring deep domain expertise	Executes under governance
System	Work triggered by events or schedules	Executes under governance
Human + AI	Work requiring collaboration	Shared authority

---

Allocation Factors

Factor	Description
Capability	Can this actor perform the work?
Role	Is this actor authorized for this work?
Capacity	Does this actor have bandwidth?
Context	Does this actor have the relevant context?
Risk	What is the risk level of this work?
Cost	What is the cost of this allocation?

---

Allocation Rules

Rule	Description
Capability-first	Work goes to the most capable actor
Role-scoped	Work stays within authorization boundaries
Risk-gated	High-risk work goes to humans
Cost-aware	Allocation considers token and compute costs
Auditable	Every allocation decision is logged

---

Human + AI Collaboration

The most powerful allocation is Human + AI working together:

Human                    AI
  │                       │
  ├── Understanding ──────┤
  ├── Decision ───────────┤
  ├── Execution ──────────┤
  ├── Verification ───────┤
  └── Learning ───────────┘

This is not "AI replaces human" or "human supervises AI." It is genuine collaboration where each contributes their strengths.

---

Work Allocation Connects to

Layer	Relationship
Work Graph	Reads work items to allocate
Human Workbench	Assigns work to humans
Twin Workbench	Assigns work to Twin
Governance	Evaluates allocation against policies
Agents	Assigns work to agents
Execution	Triggers execution after allocation