Principles Library

Entrowise's Principles Library contains 20 fundamental delivery principles organized by domain. Each principle captures a truth about how healthy delivery systems behave—and what goes wrong when it is violated. These principles form the backbone of the PPA Method (Problem → Principle → Action).

Browse by category. Each principle page includes the principle intent, early warning signs of violation, systemic consequences, coaching questions, anti-patterns to avoid, and recommended practices for making the principle a system property.

Five Diagnostic Territories

Each category targets a distinct class of system problem. Start with the territory that matches what you are facing.

Flow & Delivery Dynamics

Diagnose why work moves slowly, unpredictably, or accumulates hidden delay.

System Question: How efficiently and predictably does value move through the system?

This category contains principles that govern how work enters, moves through, and exits the delivery system. Violations produce long lead times, delivery unpredictability, firefighting, and release instability. In agentic systems, execution speed increases dramatically while coordination overhead shifts from humans to pipelines, making invisible queue buildup and downstream error amplification harder to detect.

Learning, Adaptation & Decision Quality

Diagnose whether the organization is learning fast enough to make valid decisions under uncertainty.

System Question: Is the system capable of updating itself based on evidence?

This category contains principles that govern how organizations form beliefs, test them, and change direction based on what they learn. Violations produce output without impact, strategy drift, false confidence, and metrics disconnected from reality. In agentic systems, prediction replaces observation, velocity hides decision degradation, and automation scales assumptions before they are validated.

Governance, Accountability & Decision Authority

Diagnose whether the organization can safely govern increasingly autonomous systems.

System Question: Who is allowed to decide, who is accountable, and how is control maintained?

This category contains principles that govern how authority is structured, how accountability is assigned, and how decisions are made visible and traceable. Violations produce accountability fragmentation, governance ambiguity, escalation bottlenecks, and unsafe autonomy. Most organizations treat governance as security review or model approval. Entrowise treats governance as continuous operational system design.

System Integrity & Architectural Coherence

Diagnose whether the system can maintain coherence as complexity increases.

System Question: Can the organization still understand, reason about, and safely evolve the system?

This category contains principles that govern how systems hold together as they grow, accumulate change, and add new components. Violations produce hidden dependencies, cascading failures, and an inability to explain or evolve the system safely. In agentic systems, complexity compounds faster and system coherence degrades without deliberate architectural discipline.

Human-AI Collaboration Dynamics

Diagnose whether humans and agents are reinforcing each other or degrading each other.

System Question: Is AI augmenting human judgment or replacing it prematurely?

This category does not exist in classical Agile, Lean, or Scrum. It is Entrowise's most differentiated diagnostic territory. As AI agents take on more execution, whether humans remain capable of governing what they have built becomes a first-order delivery risk. Violations produce human disengagement, blind trust in AI outputs, oversight collapse, and judgment atrophy. This category will grow as agentic delivery patterns mature.

Flow & Delivery Dynamics

• Work Intake Within System Capacity Sustains Flow — Keep the amount and rate of work entering active delivery within the system's effective capacity to absorb, review, and complete it, and make that work visible enough to manage.
• Upstream Quality Reduces Downstream Failure Propagation — Design quality into the earliest stages of delivery so defects, weak assumptions, and invalid outputs are detected or prevented before they propagate through the system. The later a quality issue is discovered, the more downstream work may already depend on it. Upstream quality controls reduce rework, instability, and the cost of correction.
• Reducing Non-Value Work Increases Value-Creating Capacity — Reduce work that consumes time, attention, coordination, or resources without meaningfully contributing to the intended outcome or protecting a necessary system constraint. Removing non-value work increases the capacity available for activities that create, validate, protect, or enable value.
• Explicit Constraints Clarify Priorities — Make the real boundaries affecting an outcome explicit so that trade-offs can be resolved and priorities can be set with confidence. Relevant constraints may include customer commitments, regulatory obligations, funding limits, technology boundaries, capacity, security, quality, or risk tolerance. When constraints remain implicit, teams make different assumptions and priorities become unstable.
• More Handoffs Increase Coordination Cost — Reduce unnecessary transfers of work between people, teams, systems, and agents. Every handoff introduces some combination of waiting, context loss, coordination effort, ownership ambiguity, and risk of misinterpretation. As handoffs multiply, these costs compound and end-to-end delivery becomes slower and harder to manage. The goal is not to eliminate all handoffs, but to make necessary ones explicit, efficient, and well-owned.

Learning, Adaptation & Decision Quality

• Evidence-Based Decisions Reduce Assumption Risk — Base important decisions on observed evidence rather than untested assumptions, prediction, opinion, or authority. Assumptions are unavoidable, but risk grows when they go untested, when evidence cannot reach decisions before they are locked in, or when a definition of value is allowed to outlive the conditions that created it. Change is not a planning failure; it is evidence that learning has occurred.
• Progressive Commitment Reduces Uncertainty Exposure — Limit the size, scope, and reach of a commitment while important uncertainty remains, then expand it as credible evidence emerges. Commit to an outcome, not to a fixed scope or output, so execution can adapt while intent stays stable.
• Continuous System Learning Sustains Performance — Continuously learn from system behavior, human and AI alike, and use that learning to improve how work is designed, delivered, governed, and supported. Delivery systems accumulate friction, outdated practices, and new failure modes as conditions change. Sustained performance depends on the system’s ability to observe, learn, adapt, and retain what works. AI-influenced decisions have to be part of that learning loop just as much as human ones; nothing exempts them from it by default.

Governance, Accountability & Decision Authority

• Making System Conditions Visible Enables Timely, Informed Decisions — Make the conditions shaping system performance visible, and build the ability to trace decisions to their causes before a system operates, not after something has already gone wrong. Decisions grounded in shared, traceable evidence produce better outcomes than decisions grounded in assumptions, filtered reporting, or records reconstructed from memory after a failure. Visibility should extend beyond progress and status to include workload, dependencies, risks, uncertainty, constraints, decision assumptions, quality signals, and emerging outcomes.
• Accountability Requires Authority to Influence Outcomes — Hold people accountable only for outcomes they have sufficient authority, access, resources, and control to meaningfully influence, and ensure that accountability always sits with a named human, even when decisions or actions are supported, recommended, or executed by AI. Automation may assist or execute, but responsibility for results cannot transfer to the system itself. Accountability becomes effective when responsibility is matched with the ability to shape the decisions that determine the result. This also requires explicit decision boundaries: when it is unclear who may recommend, decide, approve, execute, override, or reverse an action, authority drifts while responsibility remains ambiguous, or gets deflected onto the system that acted.
• Clear Operating Rules Grounded in Intent Enable Autonomous Decisions — Translate intent into clear operating rules, policies, boundaries, and decision criteria so people and AI systems can act autonomously without losing the purpose those rules are meant to serve. Too little clarity produces divergent interpretation and constant escalation. Too much prescription replaces judgment with mechanical compliance. For AI systems, precision has to be established before execution begins because agents cannot correct a vague goal through conversation the way people do mid-task.
• Excessive Choices Slow Decisions and Weaken Commitment — Maintain a manageable set of active choices so decisions can be made with sufficient clarity, speed, and ownership. More options can improve exploration, but when the number of choices exceeds the capacity to evaluate them meaningfully, decision latency increases, priorities become less clear, and commitment weakens. In AI-augmented environments, this risk increases because options, ideas, and alternatives can be generated much faster than people or governance systems can evaluate and act on them.

System Integrity & Architectural Coherence

• Optimizing the System as a Whole Improves and Sustains System Outcomes — Optimize decisions, improvements, and resources based on their effect on the overall system rather than the performance of individual teams, functions, processes, or components. Improvements in one part of a system do not necessarily improve the system as a whole and can create delays, excess work, dependencies, rework, or unintended consequences elsewhere. Sustainable system performance comes from understanding how the parts interact and evaluating local improvements against broader outcomes such as value, quality, reliability, responsiveness, and flow.
• End-to-End Capability Reduces Dependency and Improves Flow — Design the unit responsible for delivering value with enough capability, access, and context to complete most of its work without repeatedly depending on external groups. The more often work must cross organizational boundaries for routine skills, approvals, or execution, the more waiting, coordination, and handoff cost the system creates. End-to-end capability does not mean eliminating specialization or making every team fully self-sufficient. It means reducing structural dependencies that repeatedly interrupt flow and weaken ownership.
• New Solutions Create New System Constraints — Every solution introduced to address a problem creates new constraints elsewhere in the system. System health depends on anticipating and managing these second-order effects, not just solving the immediate issue.
• Understand the Original Intent Before Removing or Replacing Existing Capabilities — Before removing or replacing existing capabilities, understand why they were introduced and what risks they were designed to manage. Many systems encode historical learning that is no longer obvious but still relevant.

Human-AI Collaboration Dynamics

• Greater Observability Enables Safer Expansion of Autonomy — Ensure AI systems are observable and understandable before granting them autonomous action. Humans must be able to see, interpret, and learn from AI behavior before autonomy is expanded.
• Automate by Task Nature, Not Capability — Match the method of execution to the nature of the task, not to what AI is technically capable of attempting. Deterministic tasks, where a fixed rule reliably produces the correct output, belong in deterministic automation. Tasks that require interpretation, synthesis, or judgment may be suited to AI delegation, but only when they are also well-defined, reversible, and verifiable. Tasks that are ambiguous, carry irreversible consequence, or depend on organizational context the system cannot access should not be delegated to AI, regardless of whether AI can attempt them.
• Agent Trust Must Be Continuously Earned, Not Historically Assumed — Trust in an agent system must be grounded in current, observed behavior, not in historical performance. An agent that performed reliably in the past provides no guarantee of reliable performance today. As context shifts, configurations age, and novel situations arise, trust based on history becomes a governance liability.
• Agents Must Surface Uncertainty Explicitly — Agent systems must be designed to surface uncertainty, ambiguity, and boundary conditions rather than proceeding with false confidence. An agent operating outside its reliable range must flag it rather than execute as if the range were sufficient. The human can only intervene at the right moment if the system signals when intervention is needed.

How to Use This Library

When facing a recurring delivery problem, use the PPA Method: first understand the problem at the system level, then identify which principle from this library is being violated, then take deliberate action that reinforces the principle. Each principle page includes coaching lens questions and anti-patterns to guide this process.

Learn how principles are applied in the PPA Method | Use the AI agent to identify violated principles in your situation | Join a leadership cohort program