Designed against the PMI Standard for AI in PPPM
PMI wrote the standard.
Canopy runs it.
PMI’s Standard for Artificial Intelligence in Portfolio, Program, and Project Management defines the principles for adopting AI responsibly. The Canopy Method is the executable system that instantiates the human-oversight, governance, and decision-accountability principles in shipped, CI-enforced code — not a framework deck.
Two layers, not two rivals
Governance is a document until something enforces it. Canopy is that something — a complement to the standard, not a competitor to it.
PMI governs
- Voluntary consensus guidance — a document
- Tells human managers how to adopt AI as a supervised tool
- Sits on top of whatever method you already run
- Eight principles, unweighted; explicitly non-enforceable
Canopy runs
- Primitives enforced in shipped code, checked in CI
- A runtime where humans and AI agents do the work under one audited vocabulary
- Replaces the method — continuous Flow, no ceremonies
- A small enforced contract your build actually verifies
Principle by principle — honestly
Where Canopy operationalizes a PMI principle in shipped code, and where it deliberately declines. No principle is “fully covered” — that would not be true.
Risk (human-in-the-loop)
OperationalizesShips: A first-class approval gate, capability-scoped human-vs-agent identity, and every override recorded in an immutable audit trail with its reason and approver.
Does not: Does not detect AI bias, hallucination, or drift. The gate is enabled by policy, not on by default.
Governance & Compliance
OperationalizesShips: Approval gates, an immutable access-controlled audit trail, and a human-vs-agent authority model — surfaced in a live Approval Gate Queue and a weighted Risk Heat-Map.
Does not: Not regulatory compliance. It produces the audit substrate that feeds compliance; it does not verify against any law.
Ethics & Professional Responsibility
OperationalizesShips: Every action carries a human-or-agent identity; consequential moves can be held for a recorded human approval; who-decided-what stays traceable and reviewable.
Does not: No fairness, explainability, or privacy framework. Accountability is operational; ethics modeling is not.
Strategic Value
PartialShips: Every unit of work is an Outcome — a result that delivers value, not a unit of activity — with real lead-time and throughput so value ships every week.
Does not: Delivery is measured; business value and ROI are not quantified.
Stakeholder Engagement
PartialShips: Audience-tailored briefings — technical, business, and regulatory lenses over one source of truth — plus a live customer Portal and an AI-drafted, human-edited Weekly Pulse.
Does not: No stakeholder segmentation or AI-adoption change management.
Optimization & Innovation
PartialShips: AI generates your Signal feed, attention alerts, and a draft Pulse — a human decides — while lead-time, throughput, and decision-velocity trends make efficiency measurable.
Does not: No model optimization or self-learning loop.
Data Quality
PartialShips: Context is a first-class, living, version-traceable input — PMI's input-for-impact principle, made an entity — backed by an immutable audit trail.
Does not: No data-governance framework or model-pipeline lineage.
People & Culture
DeclinesShips: A team-adoption tool onboards people to the method's vocabulary, and the human/agent split makes roles explicit.
Does not: No leadership, literacy, or change-management framework. A deliberate blind spot.
What that looks like in the product
Human-in-the-loop, in code
AI-agent actions carry a distinct, capability-scoped identity. Consequential Outcomes can be held for explicit human approval. Every override is permanently recorded with its reason and the approver it bypassed.
Governance you can see
Approval gates, an immutable audit trail, and a human-vs-agent authority model are enforced in the system and checked by seven CI gates — surfaced in a live Approval Gate Queue and a weighted Risk Heat-Map.
Value ships continuously
Every unit of work is an Outcome — a result, not an activity. Canopy tracks real throughput and lead time with a four-week trend, so teams watch value ship every week instead of at the end of a long cycle.
Decisions are accountable
Decisions are a first-class, measured primitive and the audit trail is structurally immutable, so AI accountability is operational, not aspirational.
What Canopy is not
The standard is broad on purpose; Canopy is narrow and enforced on purpose. Canopy is the project-execution runtime for AI governance. It is not:
- An MLOps or data-governance platform
- A detector of model bias, hallucination, drift, or explainability
- Regulatory compliance, legal, or contract tooling
- A change-management or AI-literacy program
- Affiliated with, endorsed by, or certified by PMI
The complete picture
Canopy is one clan in a complete operating system
PMI organizes the world as portfolio → program → project, under governance. Nfinit Monkeys built that as a living, AI-native system — and Canopy is the execution method inside one tier of it. Every layer PMI describes is a real, running primitive.
The whole ecosystem of ventures.
A coordinated line of business.
A product or initiative team — where the Canopy Method runs.
The lead that owns a clan's outcomes.
Council governance, approvals, and orchestration — the BOSS layer.
Skill packs agents wield, published once and reused.
Parallel agents executing the work.
PMI wrote 297 pages on how to manage AI across portfolio, program, and project. Nfinit Monkeys built it — and runs on it. Canopy is the part you can adopt today; BOSS (the Business Optimization & Saving Suite) is the operating system behind it.
Complete for how we run our own work. You adopt the operating system piece by piece — Canopy is the entry point, not the whole obligation. Your legal, regulatory, and people duties stay yours; we give you the runtime and the governance to discharge them.
Canopy is the only post-Scrum methodology whose human-approval gate, capability-scoped agent identity, and immutable audit trail are shipped as executable, CI-enforced code — operationalizing the PMI Standard for AI’s principles of Risk, Governance, and Ethics at the project-execution layer.
The Canopy Method is an independent product of Nfinit Monkeys. It is not affiliated with, endorsed by, certified by, or sponsored by the Project Management Institute. PMI, PMBOK, and Disciplined Agile are trademarks of the Project Management Institute, Inc. References to PMI’s Standard for Artificial Intelligence in Portfolio, Program, and Project Management are for descriptive comparison only. Capability descriptions reflect shipped functionality as of publication.