Governance
Establishes the standards, boundaries, authority, review expectations, and controls that guide the rest of the system.
Governance FrameworkDistributed Cognitive Prompting Framework
Distributed Cognitive Prompting Framework for AI Collaboration
Organize sustained human-AI collaboration beyond the limits of a single conversation.
DCPF is an experimental human-AI collaboration architecture that organizes Governance, Research, Knowledge, and Runtime activities into coordinated responsibilities. It is designed to help people maintain an explicit role in establishing important project information, standards, accepted knowledge, and execution instructions that result in persistent, reviewable, and reusable resources across sustained AI-assisted work.
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 setting, validate the framework for your own domain and environment, and retain human judgment, verification, and accountability at every stage.
What Problem Does DCPF Address?
Conversational AI works well for many short, self-contained tasks. The challenge emerges when a useful conversation grows into a project that continues across days, weeks, months, files, research, decisions, and repeated executions.
As that work grows, people can spend increasing amounts of time managing the AI rather than advancing the project. Important instructions may need to be repeated. Previously reviewed information can become difficult to locate. Decisions may be buried in earlier conversations. Research and execution can become mixed together. Standards may be applied inconsistently, and users may have to reconstruct context before meaningful work can continue.
These are not simply AI memory problems. They are human collaboration and project-management problems created when important responsibilities remain trapped inside a conversational environment.
DCPF is designed to address those pain points by giving recurring responsibilities a durable place outside any one conversation. Governance preserves the standards and boundaries people establish. Research separates information gathering from later execution. Human-reviewed Accepted Knowledge provides a maintained foundation for future work. Runtime gives AI clear instructions for performing tasks from that foundation.
The result is an architecture intended to help people spend less time rebuilding context, repeating established instructions, rediscovering earlier decisions, and correcting work that has drifted away from the project, while keeping human judgment and acceptance visible throughout the process.
DCPF gives recurring project responsibilities a durable home outside any one AI conversation.
DCPF as a Collaboration Architecture
The Distributed Cognitive Prompting Framework for AI Collaboration is a four-framework architecture for organizing AI interactions. Rather than placing a project's research, accepted knowledge, execution instructions, and standards into one growing prompt or temporary conversation, DCPF separates those responsibilities so they can be developed, maintained, and reused.
Prompt engineering optimizes individual interactions. DCPF organizes collections of interactions into persistent architectural systems.
DCPF is not an AI model, software platform, autonomous agent, or replacement for human judgment. It can be implemented with ordinary documents and folders or with more advanced repositories and tools. The architecture remains stable while the implementation can adapt to the project.
The Four DCPF Frameworks
Research
Directs how information is sought, sourced, evaluated, documented, and prepared for human review.
Research FrameworkKnowledge
Preserves human-reviewed Accepted Knowledge in a maintained Knowledge Repository so it can be reused without rebuilding the project from conversation history.
Knowledge Framework and RepositoryRuntime
Uses approved Knowledge and current execution instructions to perform a defined task and produce a Project Output without quietly reopening Research or changing the repository.
Runtime FrameworkHow DCPF Works
Human Defines the Work → Builds the Relevant Artifact With Applicable Governance Controls
→ AI Researches or Executes → Human Review / Acceptance → Accepted Knowledge Is Preserved
When Applicable → Feedback / Update the Appropriate Framework as Needed
Governance applies across the system rather than functioning as a one-time first step. It defines the standards and boundaries that shape Research, control what may become Accepted Knowledge, guide Runtime, and establish review and approval expectations.
AI can research, organize, interpret, draft, compare, transform, and execute within the structure supplied to it. The human collaborator determines what the project should do, what information it should rely upon, what becomes Accepted Knowledge, what outputs are accepted, and whether those outputs are appropriate to use.
Outputs also provide feedback. Missing information may point back to Research, outdated information to Knowledge, execution problems to Runtime, and recurring cross-project problems to Governance.
Do I Need DCPF?
DCPF should not be added merely because AI is involved.
Ordinary AI Interaction
A quick question, simple explanation, disposable brainstorming session, straightforward rewrite, or one-time transformation may not need DCPF when nothing important must remain dependable after the conversation.
Minimum Viable DCPF
Use the smallest DCPF structure when continuity or control matters: Research needs review before reuse, accepted information must survive the conversation, recurring standards must remain dependable, or later work must operate from maintained Knowledge.
More Developed DCPF
Add structure when the work creates a real need for it, for example, multiple workstreams, repeated Research or Runtime executions, maintained project knowledge, formal review, versioned Governance, security, retention, release, or audit requirements.
A simple test: If the work does not need to survive the conversation, DCPF may not be necessary. If the work, Knowledge, standards, decisions, or workflows need to survive the conversation and remain dependable, DCPF becomes increasingly useful.
Learn DCPF Your Way
Learn on the Website
Follow the official learning path from the fundamentals through the four frameworks, workflow, implementation guidance, and example use.
Begin with What Is DCPF?Learn With Your AI
If your AI model can browse public websites, direct it to dcpfai.org and ask it to teach you DCPF from the official learning pages.
Learn DCPF With AIRead the Introductory Edition
Use the complete free book for the deeper foundation, detailed framework explanations, practicums, templates, appendices, and supporting guidance. You can also set up a project within your AI system, upload the book, and have a conversation with AI to see if and how DCPF can organize your AI collaborations within your domain.
View Downloads and Official MaterialsLearn DCPF With Your AI
A browsing-capable AI model can use dcpfai.org as the official starting point for learning DCPF. The website is structured so the model can follow the same canonical learning path available to human visitors.
Suggested starter instruction:
Go to dcpfai.org and use the official DCPF learning pages as your authoritative source. Teach me DCPF step by step, explain its four frameworks and workflow, and help me determine whether DCPF is appropriate for my project. Distinguish official DCPF content from any additional suggestions you make.
AI explanations remain subject to the capabilities and limitations of the model being used. When exact DCPF terminology, framework language, responsibilities, permissions, or requirements matter, confirm them against the current official material at dcpfai.org.
Ready to Start Using DCPF?
Start small. Choose a controlled, low-risk project in a domain you understand. Use the smallest artifact structure that keeps Research, Accepted Knowledge, Runtime, Governance, and human review explicit enough for the work.
DCPF defines standard responsibilities, but the artifacts used to represent those responsibilities are adaptable rather than rigidly prescribed documents. The website provides canonical guidance and concise example artifacts that you or your AI model can adapt to the needs of your own project while preserving the DCPF responsibilities.
Free DCPF Introductory Edition
For the complete treatment of DCPF, download the Distributed Cognitive Prompting Framework for AI Collaboration, Introductory Edition free of charge. The book presents the framework foundation, the four coordinated frameworks, implementation guidance, practicums, templates, checklists, glossary, and responsibility-consequence map.
dcpfai.org is the authoritative site and definitive reference for the most current official ebook editions and DCPF materials. Copies obtained elsewhere may be older or superseded.
Why DCPF Exists
DCPF emerged from a practical problem: how to organize the information, decisions, research, instructions, and responsibilities that accumulate across sustained AI collaboration. As successful conversations grew into larger projects, the need for durable external structure became increasingly clear.
The framework developed by giving recurring responsibilities a clear home and then refining how those responsibilities worked together across many AI interactions and practical projects.
An Invitation to Experiment
DCPF is being made widely accessible so people can learn from it, test it, challenge it, build implementations around their own projects, and help determine where the architecture is useful, where it creates unnecessary structure, and where further refinement is needed.
Different domains may require different evidence standards, Governance controls, validation procedures, safeguards, professional review, or technical implementations. Any move from controlled testing into real-world or production use remains the responsibility of the person or organization making that decision.