The Discipline · Architecture

The architecture.

The working diagram of how a HI-Centric-AI system is structured. Practitioners enter from the left. The model substrate, top left, routes the request through inference. The knowledge layer, top right, binds curated, audited, retrieval-grounded material. At the structural center sits HI-CENTRIC-AI itself — the keystone node through which every articulated decision passes. Articulated by Daniel William Dorsey and Daniel Nowak.

ATUMNUS

ARTIFICIAL INTELLIGENCE

SERVER

ATUMNUS

APPLIED INTELLIGENCE

ARCHITECTURE

ATUMNUSAPPLIED INTELLIGENCEINFRASTRUCTURE

NEW DATAVARIABLE INJECTION

ATUMNUSAPPLIED INTELLIGENCEINFRASTRUCTURE

Extension

Knowledge Base

Tool

Web Search

Data

User History

DATA VARIABLE

ATUMNUSAPPLIED INTELLIGENCEINFRASTRUCTURE

HI-CENTRIC-AI

—HI—
Human Intelligence

Merge

broad reference

Vector Storage

real-time retrieval

Safety Check

session context

Check Context

semantic search

Personalize

relationship mapping

Output

compliance validation

KNOWLEDGEOUTPUT

KNOWLEDGEPURPOSE

Articulated by Daniel William Dorsey · Daniel Nowak

How to read the diagram

HI-CENTRIC-AI sits at the structural center.

The diagram resolves into two halves that meet at the keystone. The substrate — inference, routing, generative reach — is the AI. The keystone — named authority, curated knowledge, audit trail — is the Hi. Each is necessary; neither is sufficient.

Half I · The substrate

What inference brings.

Top left of the diagram. The model layer routes every request through inference, generative completion, and semantic retrieval. This is the AI in HI-Centric-AI — a substrate of compute, weights, and sampling. Necessary, but not authoritative on its own.

Model routing
Requests dispatched to appropriate inference endpoints based on task class.
Generative reach
Open-ended completion across domains, voices, and formats.
Retrieval grounding
Semantic search across the bound knowledge layer; no fabrication.
Substrate trail
Every inference step logged for downstream attribution.
Compute provisioning
Token budgets and latency bounds set by the operating signatory.
Capability ceiling
Inference can extend, but cannot authorize.

Half II · The keystone

What HI brings.

Structural center of the diagram. The keystone node holds named authority, curated knowledge, and audit trail. This is the Hi in HI-Centric-AI — every articulated decision passes through it. The keystone authorizes; the substrate executes.

The founding principles

Named authority
Every decision signed by an identified practitioner; no anonymous outputs.
Curated knowledge
Provenance, voice, and lineage are tracked at the document level.
Audit trail
The accumulating record of who decided, citing what, under what mandate.
Knowledge Output
The articulated result of the architecture — the right side of the diagram.
Knowledge Purpose
The named intent under which the output was produced.
Compounding asset
Each decision enriches the named-authority archive in operation.