Distributed
Research, preserved Knowledge, execution, and Governance have distinct responsibilities rather than being crowded together.
What Is DCPF?
A collaboration architecture for sustained human-AI work.
A Collaboration Architecture for Sustained AI Work
The Distributed Cognitive Prompting Framework for AI Collaboration, or DCPF, is an experimental human-AI collaboration architecture for organizing sustained work across four coordinated responsibilities: Governance, Research, Knowledge, and Runtime.
Instead of expecting one prompt or one conversation to carry a project's evidence, accepted information, instructions, standards, decisions, and execution history, DCPF gives those responsibilities distinct places where they can be developed, reviewed, maintained, and reused.
The human directs the system. The architecture organizes the work. The AI reasons within that structure.
DCPF is experimental.
DCPF has demonstrated promise through its development work, practical applications, and training practicums, but it has not been broadly or independently validated across domains, organizations, AI systems, or production environments. Begin in a controlled, low-risk testing environment, validate your own implementation for your domain and AI environment, and retain human judgment, verification, and accountability throughout the process.
Why DCPF Exists
DCPF emerged from a practical problem. As useful AI conversations grew into sustained projects, the amount of research, accepted information, decisions, instructions, standards, revisions, and repeated work also grew. Information that had been easy to manage inside a conversation became harder to preserve, distinguish, reuse, and govern.
Prompt-centered work can produce excellent individual responses. Sustained projects introduce a different problem: important context can become buried in conversations; approved and unapproved material can blur together; standards may need to be repeated; research, instructions, and outputs can become mixed; and one prompt may be expected to carry too many responsibilities.
DCPF responds by moving recurring organizational responsibilities into an external architecture. The goal is not simply to store more information. It is to organize information by responsibility while keeping the parts that depend on one another connected.
What Does “Distributed Cognitive Prompting” Mean?
DCPF uses the phrase Distributed Cognitive Prompting to describe a way of spreading important cognitive responsibilities across coordinated structures instead of concentrating them inside one prompt or conversation.
Cognitive
The architecture supports reasoning, learning, organization, judgment, and decision-making, not text generation alone.
Prompting
Prompts remain essential for directing work, but they operate inside a larger maintained collaboration environment.
Distributed Cognitive Prompting, or DCP, is the broader organizational approach interpreted and applied in the book. DCPF is the specific four-framework architecture that integrates Governance, Research, Knowledge, and Runtime as a unified strategy for AI-assisted projects.
The Four Coordinated DCPF Responsibilities
DCPF separates four responsibilities that frequently become mixed together in sustained AI work. Each responsibility has a different job, but none is meant to operate in isolation.
Governance
Governance supplies human direction across the architecture. It establishes authority, boundaries, quality criteria, evidence expectations, review points, approval conditions, and other controls. Governance is active across Research, Knowledge, Runtime, and output evaluation; it is not merely a final inspection step.
Governance FrameworkResearch
Research turns an information need into a deliberate investigation. It defines what must be learned, which evidence may be used, how conflicts and uncertainty are handled, and how findings are documented for human review.
Research FrameworkKnowledge
The Knowledge Framework owns the responsibility for preserving Accepted Knowledge so it can be maintained and reused. That Knowledge is represented and maintained in the Knowledge Repository as reusable project intelligence rather than being left buried in conversation history.
Knowledge Framework and RepositoryRuntime
Runtime directs execution. It applies approved Knowledge to a current objective, identifies the Knowledge needed for the task, establishes the workflow, and defines the expected Project Output without quietly reopening Research or changing the Knowledge Repository.
Runtime Framework
Research discovers. Knowledge preserves. Runtime executes. Governance guides.
Human Direction Is Part of the Architecture
DCPF does not place the AI model in charge of the larger collaboration environment. People determine what the project is trying to accomplish, which evidence is acceptable, what Knowledge should persist, which standards apply, and what outcomes meet the required level of quality.
Inside an interaction, an AI model can reason, compare, synthesize, generate, organize, and apply the context it has been given. Outside the model, people establish purpose, preserve Accepted Knowledge, maintain sources and standards, define authority, and decide which changes should become authoritative.
The AI can help develop and evaluate these resources. It does not independently decide that a draft policy now governs the project, that one conflicting source should replace another, that new information has become Accepted Knowledge, or that its own suggestion has been approved.
Why Put Important Responsibilities Outside the Conversation?
Human beings have long extended their thinking by preserving important knowledge outside individual memory: plans, checklists, laboratory records, policies, procedures, and technical standards. DCPF applies the same principle to sustained AI collaboration.
Project goals, Accepted Knowledge, sources, workflow instructions, standards, version status, and approval decisions can exist in maintained resources that people can inspect and update. The AI receives the material required for the current task while the larger project foundation remains available for later work.
This separation lets Knowledge endure while execution adapts. Research can expand without rewriting every operating prompt. A recurring standard can be updated without changing factual Knowledge. A new Runtime task can reuse maintained Knowledge without reconstructing the project from conversation history.
The result is not less information. It is information organized by purpose.
DCPF and Prompt Engineering
DCPF does not replace prompt engineering. It gives prompting a focused role inside a larger collaboration architecture.
A prompt can direct Research, apply maintained Knowledge, start a Runtime task, or request evaluation without carrying the project's full history, all supporting material, every standard, and every responsibility.
Prompt engineering optimizes individual interactions. DCPF organizes collections of interactions into persistent architectural systems.
The unit of design therefore changes. Instead of trying to create one perfect prompt that does everything, DCPF organizes the collaboration environment in which evidence, maintained Knowledge, execution, Governance, human review, and outputs work together.
What DCPF Is Not
Not an AI model or software platform
DCPF does not operate, retrain, or alter the underlying AI system. It can be used with different models, platforms, and tools.
Not an autonomous agent
DCPF does not give the AI authority to decide what becomes authoritative, approve its own changes, or replace human judgment.
Not one enormous prompt
Separating Research, Knowledge, Runtime, and Governance is one of the reasons the architecture exists.
Not a rigid file format
DCPF defines standard responsibilities. The artifacts and technical tools used to represent those responsibilities can be adapted to the project.
Not a guarantee of accurate AI output
A more organized project can make work more structured and reviewable, but it cannot eliminate model errors, outdated information, bias, inconsistency, or other AI risks.
Stable Architecture, Flexible Implementation
DCPF is designed so the architecture can remain recognizable while implementation changes with the work. A small personal project may use a few coordinated documents and a basic folder. A larger project may use maintained repositories, versioned Governance, templates, specialized workflows, databases, applications, or automation when the project creates a real need for them.
Governance is an architectural responsibility from the beginning, but a separate Governance Repository is not required before a beginner can start. Use the Governance Framework responsibilities while building Research, Knowledge, Runtime, and review artifacts, and state the applicable controls directly where they operate. As recurring controls emerge, preserve them in the Governance Repository for consistency, reuse, versioning, and oversight.
The Governance Framework guides. Building artifacts reveal additional Governance controls. The Governance Repository preserves recurring Governance.
The architecture grows because the work requires it, not because complexity appears impressive.
DCPF Organizes the Structure; You Control the Contents
DCPF can be understood in part like a structured form or a filing cabinet. The structure provides fields, categories, folders, labels, and relationships, but it does not author, verify, approve, or own the information placed inside it.
Users decide what information and instructions to supply, which AI systems and tools to use, how to represent DCPF responsibilities, what outputs to accept or reject, and how their implementation will be tested and maintained. Subject to applicable law and the rights of others, users retain ownership of content they author and remain responsible for the content they create, select, adapt, or incorporate.
DCPF in One View
- DCPF is an experimental architecture for sustained human-AI collaboration.
- It distributes responsibility across Governance, Research, Knowledge, and Runtime.
- Research develops evidence; humans review and accept what may become maintained Knowledge.
- The Knowledge Repository preserves Accepted Knowledge for reuse.
- Runtime applies approved Knowledge to current work.
- Governance carries human-authorized standards and controls across the architecture.
- The AI reasons within supplied context; people retain authority over purpose, acceptance, standards, and use.
- DCPF complements prompt engineering rather than replacing it.
- The architecture is stable; the files, tools, and technical implementation can adapt to the project.
- Because DCPF is experimental, users should begin with controlled, low-risk testing and validate it for their own environment.
Where to Go Next
Now that you know what DCPF is, the next step is to see how information and human decisions move through the architecture: from Research, through human review and Accepted Knowledge, into Runtime and Project Outputs, with Governance operating across the system.