Decision Architecture
Research at Fragment Practice is where observations from real work are turned into concepts, language, and reusable structures.
If Services is where operating models are shaped in practice, Research is where those recurring patterns are named, clarified, and developed into the building blocks of Decision Architecture.
This includes field notes, conceptual essays, practical frameworks, maps, patterns, and applied studies that connect human judgment, organizational structure, and AI-enabled work.
Looking for practical engagement formats? See Services (JP). Looking for the broader studio context? See About.
Fragment Practice runs on a simple loop: observe recurring decision problems in real work, abstract them into concepts and frameworks, then return them to practice in a reusable form.
The research currently develops across four connected lines.
How decisions actually move through meetings, notes, logs, approvals, handovers, and AI-supported workflows.
How review, escalation, documentation, and policy interact with AI-enabled decisions in practice.
The internal language of the field: maps, principles, canvases, and patterns that make Decision Architecture communicable and practical.
Essays, ZINEs, and applied notes that connect everyday life, work, institutions, and AI through the structure of decisions.
These are not “theories” in the abstract. They are working layers for deciding what to structure, where to draw boundaries, and how to keep systems legible.
A central research layer focused on where authority shifts between actors — humans, teams, organizations, and AI systems.
Research into how decisions can remain readable, reviewable, and auditable after the moment of action.
How difficult cases move through systems, and how review loops transform individual decisions into organizational learning.
Research into the gradual movement from AI assistance to AI default authority — often without explicit redesign of responsibility.
Research begins small. The goal is not scale first, but legibility first.
Start from a real decision problem in practice: repeated ambiguity, missing trails, unclear ownership, or unsafe acceleration.
Translate recurring problems into concepts, distinctions, or design patterns that can travel beyond the original case.
Bring those concepts back into practice through advisory work, writing, and public language that others can reuse.
Research is published in forms that remain readable and reusable.
Research is the conceptual backbone. ZINE is often the most accessible entry point. Services is where these ideas are translated into operating models.
