Agent workforce architecture

Agent Workforce Architecture

Overview

Use this skill to support the leader as AI workforce architect in a mega-manager operating model. A clear operating model for what humans own, what agents own, where handoffs happen, and how the whole system is governed.

A mega manager is not a person who passively supervises more humans. It is a leader who manages a portfolio of humans, AI agents, workflows, memory, tools, evals, and approval gates. The agent expands span of control only when the operating system is legible, governed, and reviewable.

When to Use

Run this skill when:

• Leader wants to scale output with AI agents or AI employees
• Team is using many tools/agents without a coherent operating model
• Work needs to be decomposed into human, agent, and human-plus responsibilities

Do not use this skill to bypass judgment, accountability, security, privacy, HR, legal, customer approval, or executive decision rights.

Inputs

Gather:

• Business outcomes and recurring work inventory
• Existing team roles, systems, tools, and data sources
• Risk classes, approval boundaries, and compliance constraints
• Current agent capabilities and failure modes

If key inputs are missing, label assumptions and confidence. Do not invent tools, access, facts, policies, or authority.

Workflow

Follow this sequence:

1. Inventory recurring work and classify by judgment, risk, repeatability, and data availability
2. Assign each work class to human, agent, or human-plus ownership
3. Define autonomy levels from draft-only to supervised execution to bounded autonomous operation
4. Specify tool/data access, memory boundaries, and audit logs for each agent role
5. Create operating cadence, escalation paths, and kill switches

Always finish by making the control loop visible: owner, current state, review point, approval boundary, and kill/rollback rule where relevant.

Output Format

Use this structure:

# Agent Workforce Architecture

## Objective
[What system, workflow, agent, or team capability is being designed or reviewed.]

## Current State
- Humans:
- Agents/workflows:
- Tools/data:
- Risks/unknowns:

## Design or Review
[The architecture, brief, review, command center, governance plan, eval suite, or backlog.]

## Autonomy and Approval Boundaries
- Agent may:
- Agent must not:
- Human approval required for:

## Verification
- Acceptance criteria:
- Evidence required:
- Review cadence:
- Kill/rollback trigger:

Expected deliverables:

• Human-plus-agent operating model
• Agent role catalogue
• Autonomy and approval matrix
• Escalation map
• Implementation roadmap

See assets/output-template.md for a reusable version.

Human Decision Boundary

The agent may prepare, structure, evaluate, monitor, and recommend. The human leader owns final decisions, accountability, and risk acceptance. The agent must not cross these boundaries:

• Do not grant agents access to sensitive systems without explicit approval
• Do not automate public, legal, financial, HR, or customer commitments
• Human accountable leader owns operating model and risk acceptance

Stop for explicit approval before granting access, increasing autonomy, sending external messages, making people/customer/financial/legal commitments, changing production systems, or retaining sensitive memory.

Quality Bar

A strong output for this skill:

• Makes the human-agent operating model more legible, not more magical.
• Names owner, state, authority, review cadence, and failure response.
• Uses evidence and acceptance criteria instead of vibes.
• Reduces managerial drag without eroding accountability.
• Includes safety boundaries appropriate to autonomy level and data sensitivity.
• Creates reusable artifacts a leader can run repeatedly.

Failure Modes

Watch for these mistakes:

• Treating agents as employees with intent instead of systems with failure modes.
• Scaling autonomy before evals, logging, approval gates, and rollback exist.
• Creating invisible work that nobody owns or reviews.
• Confusing polished output with verified output.
• Adding more agents when the real problem is unclear workflow ownership.

References

• Anthropic: Building effective agents, workflow and agent patterns
• OpenAI agentic workflow guidance and eval emphasis
• Ethan Mollick, Co-Intelligence: human judgment with AI collaboration
• Microsoft Work Trend Index: AI reshaping knowledge work and management

For the shared methodology spine, see ../../docs/SOURCE-SPINE.md.