One model, six layers
Institutional Intelligence Architecture is not a product diagram — it is a reference architecture: a layered model that any AI-native organization can adopt regardless of industry, deployment, or vendor. It expresses a single idea — that reasoning, not just data, is the asset worth preserving — as six cooperating layers.
Read top to bottom, the layers move from the decisions an organization is accountable for down to the infrastructure those decisions ultimately run on. Each exists to make the layer above it more capable and more defensible.
The layered stack
Hover a layer to inspect its responsibility; select it to open the full layer page.
How the layers relate
Three principles govern how the six layers fit together into one coherent architecture:
Single responsibility
Each layer does one thing. Business owns outcomes; Knowledge owns memory; the Intelligence Layer owns reasoning; Application owns surfaces; Technology owns runtime; Infrastructure owns environment.
Every layer reasons over the one beneath it
Higher layers are projections of lower ones. Applications are projections of the intelligence and knowledge layers, not separate copies — so nothing drifts out of sync.
Substitution without rewrite
Change the Infrastructure layer from cloud to air-gapped, or swap a model in the Technology layer, and the layers above are unaffected. Trust posture never forces an application rewrite.
Explore each layer
Each layer has a dedicated reference page. Follow them in order, or jump to the one you need.