Open standard · CC BY 4.0
D-POAF - Decentralized Proof-Oriented AI Framework
An open framework for AI-Native software engineering
where every AI contribution is verifiable, traceable,
and accountable by design.
DEEP INTENT
AI-native engineering can't be justified by hierarchy or performance alone.
AI-native engineering introduces decisions and changes that cannot be justified by hierarchy, central control, or performance claims alone. In hybrid human-AI environments, trust, value, and responsibility require decision authority to be distributed, governed through evidence, and demonstrable through traceable proofs over time. D-POAF establishes proof-first legitimacy as the foundation of engineering, governance, and system evolution.
What D-POAF enables
- Proof-first legitimacy for every engineering decision
- Distributed authority across human-AI environments
- End-to-end traceability from intent to outcome
THE WAVE LIFECYCLE CORE
Intent → Decision → Execution → Evidence → Learning → Adaptation
Four phases. One Wave. Continuous Proof.
D-POAF structures system evolution as Waves. Each Wave traverses four implementation-independent macro-phases to produce and refresh verifiable proof and to update the operating envelope for the next Wave. A Wave is the unit of verifiable progress.
Phase 1
Instruct & Scope
Translate intent into a scoped instruction grounded in context, with explicit scope boundaries, initial hypotheses, and proof expectations.
Phase 2
Shape & Align
Refine and decompose scope, align decision and investment logic, and prepare execution through prompt-action design and human ↔ AI delegation bounds.
Phase 3
Execute & Evolve
Execute prompt actions, validate delivery (PoD), review outcomes (PoV), and monitor reliability (PoR) producing and refreshing proofs over time.
Phase 4
Learn & Adapt
Close and re-open the loop. Synthesize outcomes, capture lessons, and update intent, governance, and standards for the next Wave.
Eight domains. One shared framework.
D-POAF structures the AI-native engineering ecosystem into eight functional domains. Each domain has its own scope, its own proof obligations, and its own governance boundaries, all grounded in the same Wave Lifecycle and Proof Model.
Open a domain to explore its scope and intended outcome.
AI Development & Engineering
Build AI-native software with governance and traceability by design.
- Human–AI collaboration workflows
- Prompt and code traceability
- Integration with development tools: Git, CI and more
- Verifiable delivery: Proof of Delivery
Reliable, auditable, higher-quality development.
AI FinOps
Govern AI resources, costs and performance across the organisation.
- Usage observation and allocation
- Model selection and routing
- Cost, quality and performance optimisation
- Budgeting and forecasting
- Sovereignty and data residency
Optimised, sovereign and sustainable AI resource usage.
AI Strategy & Portfolio
Align AI initiatives with business goals and measurable value.
- Strategic planning and use case prioritisation
- Portfolio and value tracking
- Business and technical alignment
- ROI and impact measurement: Proof of Value
Focused investments and measurable impact.
AI Enablement & Adoption
Empower people and teams to work effectively with AI.
- Roles and responsibilities: RAGer, Wave Surfer and more
- Training and certification
- Guides, templates and best practices
- Change management and community
Stronger adoption, skills and responsible use.
AI Governance & Risk
Guide responsible, secure and compliant AI use at scale.
- Policies and guardrails
- Risk and criticality management
- Incident and refusal handling
- Regulatory alignment: EU AI Act, NIST, ISO
Safe, compliant and trustworthy AI adoption.
AI Operations
Operate, monitor and continuously improve AI systems. LLMOps / AIOps.
- Model and agent lifecycle management
- Monitoring, evaluation and feedback
- Continuous assurance and evidence
- Incident response and recovery
- Scalability and reliability
Robust, performant and continuously improving AI systems.
AI Security & Sovereignty
Protect data, models and operations while preserving sovereignty.
- Data classification and handling
- Sovereign model routing: local, private, public
- Access control and identity
- Security by design
- Audit and evidence
Protected, sovereign and accountable AI operations.
AI Innovation & Research
Advance the state of the art towards an open and trusted AI future.
- Research programmes and publications
- Scientific validation and benchmarks
- Collaboration with academia and industry
- Standardisation and ecosystem partnerships
Continuous innovation and real-world contribution.
Common foundations
A shared language. A continuous evidence trail.
- Canonical principles, lifecycle and proof
- Shared terminology
- Evidence and traceability
- Platform reference implementation
- Specification and conformance
- Living governance
- Standards and regulatory mappings
- Operating guides and methods
THE PROOF MODEL
Three families of verifiable evidence.
D-POAF defines three families of proof. Each answers a
different question, together they form the continuous
evidence that grounds every D-POAF decision.
Proof of Delivery (PoD)
Evidence of intended behavior and technical alignment. Did the artifact meet its spécifications and design constraints?
Proof of Value (PoV)
Evidence of outcomes and measurable impact. Did the work create the value it was meant to create?
Proof of Reliability (PoR)
Evidence of sustained quality, safety, and stability over time. Does the result hold up beyond release?
JOIN US
Adopt the D-POAF framework
D-POAF is open and free to use under CC BY 4.0.
Read the Canonical Specification, explore the docs, Big Picture Map, Reference Map
or join the community on d-poaf.org.