PAGE_ROLE: SUPPORTING IDEA MAP

Ideas

A supporting idea map for the AI Agent Lifecycle mainline. Canonical definitions live in Concepts; the main theory route is Lifecycle.

START_HERE: IDEAS_TO_ESSAY_TO_PROOF

What concepts make the thesis legible? This is a supporting idea map, not the main theory route. The main route is AI Agent Lifecycle; canonical definitions live in Concepts; essays and proof path continue the argument.

Agentic Delivery names the outcome thesis. AI Agent Lifecycle defines the accountable lifecycle of agent work. MPLP is the lifecycle protocol path for the concrete proof path.
Agent Activation LifecycleEngineering Process LifecycleProject LifecycleEnterprise Cross-Project Lifecycle
PILLAR_ID: AI_AGENT_LIFECYCLE
PILLAR_NAME

AI Agent Lifecycle

The field-definition layer for accountable agent work: how intent, context, decisions, evidence, learning, and accepted outcome stay coherent through time.

SPECIFICATION_FOCUS

What must stay continuous for agent work to remain traceable, reversible, accountable, and acceptable?

PILLAR_ID: PROJECT_LIFECYCLE_VS_TASK_EXECUTION
PILLAR_NAME

Project Lifecycle vs Task Execution

The distinction between completing a run and preserving intent, constraints, evidence, and acceptance over time.

SPECIFICATION_FOCUS

Is the system optimizing the run, or preserving the work?

PILLAR_ID: LIFECYCLE_GOVERNANCE
PILLAR_NAME

Lifecycle Governance

The missing governance layer above tool access and agent coordination: authority, constraints, confirmation, evidence, and closure.

SPECIFICATION_FOCUS

Who governs the lifecycle before, during, and after agent action?

PILLAR_ID: EVIDENCE_/_ACCEPTED_OUTCOME
PILLAR_NAME

Evidence / Accepted Outcome

A delivery claim is not complete until evidence can be inspected and the outcome can be accepted against intent.

SPECIFICATION_FOCUS

What proves that the work stayed legitimate and finished correctly?

PILLAR_ID: ACCOUNTABLE_WORK
PILLAR_NAME

Accountable Work

Work that remains tied to intent, authority, responsibility, evidence, and review instead of ending as an unaudited output.

SPECIFICATION_FOCUS

Can someone reconstruct why the agent acted and why the result should be trusted?