5. Foundational Assumptions

POOKA is based on several architectural assumptions. These assumptions are not presented as universally proven truths but as the foundation upon which the architecture has been developed.

Information outlives interaction

Conversations are temporary. Information is persistent. Sustainable Human–AI collaboration requires knowledge to exist independently of individual conversations.

Meaning should be explicit

Information becomes more reusable when its semantics are represented explicitly rather than being inferred from conversational context.

Context is architecture

Context should not primarily be reconstructed through prompts. It should be represented as a persistent architectural concept.

Governance is part of information

Questions such as who may act, on whose behalf, under which conditions and within which boundaries belong to the architecture itself rather than to implementation logic.

Structure precedes automation

AI should operate on well-defined structures instead of compensating for missing structure through increasingly complex prompts.

Human judgement remains external

AI may analyse, suggest, summarize, transform and reason. Responsibility for judgement and decision-making remains outside the AI itself.

Architectures should survive implementations

Architectural concepts should remain applicable regardless of future AI models, software platforms or storage technologies.