A formal seven-tier taxonomy of the AI technology stack is presented for enterprise architects, spanning from deterministic Symbolic AI and Knowledge Engineering at the base, through Machine Learning, Neural Networks, Deep Learning architectures, Generative Foundation Models, Agentic AI systems, to Multi-Agent Orchestration at the apex. Each tier is analyzed with mathematical rigor, covering inter-layer dependencies, governance requirements, and architectural failure modes. The paper argues that enterprise AI is a full-stack engineering discipline requiring deterministic governance anchoring probabilistic generative layers. A critical section identifies five irreducible failure modes of transformer-based systems — including hallucination, absence of causal reasoning, stateless epistemics, metacognitive opacity, and distributional brittleness — and argues that agentic scaffolding alone cannot close the resulting intelligence gap, pointing toward Metacognitive AI as the necessary next architectural paradigm.
