See the operating model.
Understand how intent, governance, execution, and learning stay connected.
Start with the overviewThe reference architecture for the Agentic PMO.
A governed operating model for turning organizational intent into coordinated execution with humans and AI.
The Agentic PMO operating loop
Organizational intent
Governed activities
Human + AI execution
Evidence + change
Living governed state
Workspace projections
Human judgment
Validated learning
Digital DNA
return to intent · continuously coordinated
Nine-node continuous loop · public release v1.0.1
Study the architecture.
Explore canonical concepts, responsibility boundaries, integration, and conformance.
Open the architectureBuild within the boundaries.
Design agents that preserve authority, evidence, explainability, and human control.
See the guardrailsOne architecture. Many implementations.
PMO City defines the operating model before the technology. The architecture is designed to remain meaningful across products, models, workflows, and enterprise systems.
Human judgment is sovereign.
AI prepares, coordinates, explains, and acts within bounded authority. Accountability stays visible.
Operational reality is governed.
Humans and AI contribute to one Living Governed State rather than disconnected semantic models.
Evidence is not automatic truth.
Observations, accepted state, uncertainty, outcomes, and enduring knowledge remain distinct.
Technology is replaceable.
Models, providers, connectors, and deployment choices realize the architecture without defining it.
Read at the depth you need.
The public publication starts with a clear map, then opens into the concepts and responsibility boundaries needed for implementation and review.
Intent, Activities, Projects, and State.
Understand the vocabulary that keeps organizational meaning stable across representations and systems.
AI with an explicit boundary.
Assignments, authority, context, tools, evidence, explainability, oversight, and safe degradation.
Native, federated, delegated.
Integrate with existing systems while preserving source authority, accountability, and reconciliation.
The public release is the starting point. It gives leaders, architects, and builders a shared vocabulary without requiring a 700-page read.
See it in practice. Follow the mandate-to-learning example before diving into definitions and conformance.
04 / See it in practice
From mandate to learning.
A fictional release-readiness scenario makes the loop concrete: intent becomes governed work, evidence changes the state, a human decides, and validated learning improves the next cycle.
A real decision, not a demo.
Leadership sets a bounded mandate with an accountable owner, a scope, and an explicit human approval boundary.
Nine steps stay connected.
See how an Activity carries intent through execution, Evidence, accepted state, judgment, and learning.
Read the worked exampleBoundaries remain visible.
The example distinguishes AI preparation from human approval, observation from accepted state, and learning from universal truth.
Review the claim boundaryStart with the question you have.
Agentic PMO is a reference architecture, not a prescription for one stack. Use the path that matches your current decision.
“What changes for my PMO?”
Begin with the definition, operating loop, and human–AI collaboration model.
Read the operating model“How do we build it?”
Use the canonical model, responsibility boundaries, integration modes, and runtime principles.
Study the architecture“How do we evaluate it?”
Review the claim boundary as a scoped question supported by evidence, not a feature label.
Review the claim boundaryA reference architecture stays open to challenge.
PMO City maintains the public release and accepts specific questions, corrections, examples, accessibility improvements, and evidence requests through the public repository.