AI-Augmented Impact Frames: A Closed-Loop Architecture for Purpose-Aligned Decisions

How to scale decision-making with AI without losing control of meaning.

AI can accelerate decisions—but also obscure why they are made. This paper introduces a closed-loop architecture that preserves human interpretive control while enabling AI-supported analysis. By constraining how meaning is translated into decisions, it prevents drift from becoming invisible inside automated systems. The goal is not just faster decisions—but traceable ones.

About the Coherence Programme
The Coherence Programme studies why institutions drift despite appearing aligned. It shows that decisions are made not on intent itself, but on how intent is translated into criteria, metrics, and allocation rules. Using the Operating Spine, the programme traces how purpose becomes action across governance layers, making drift and coherence directly observable within decision systems. The research applies to public institutions, capital allocation, and AI-mediated environments, where the durability of decision rules determines long-term institutional reliability.
Programme citation: Mertens, R. E. U. (2026). The Coherence Programme: A Conceptual Overview and Entry Point to the Research Programme.
Resources:  Coherence Programme OSF repository and https://thecoherenceprogramme.org

Version 1.00 — Initial Release:
This version presents the first formal specification of the AI-Augmented Impact Frames architecture. It defines the Operating Spine as the structural unit of analysis, specifies boundary conditions under which institutional learning can be studied, and establishes the constraints governing AI’s interpretive role.This release introduces the core architectural vocabulary and analytical commitments of the programme. Subsequent versions may harmonise terminology across related papers but will not alter the structural architecture, boundary conditions, or theoretical claims introduced here.
Version 1.01: Programme Consolidation Update: This version consolidates the manuscript within the unified Coherence Programme structure.Titles, terminology, and internal cross-references have been harmonised across the series to stabilise the programme’s core constructs:

• Translation Drift (mechanism)
• Translation Coherence (metric)
• Interpretive Maintenance (governance function)
• Distributed Coherence (theoretical integration)

No changes have been made to the formal architecture, boundary conditions, methodological logic, or theoretical claims.
The update improves cross-paper traceability, indexing consistency, and conceptual coherence across the programme.
Version 1.02: Terminology harmonisation and minor structural refinements to improve consistency across the Coherence Programme. No changes to the theoretical framework, constructs, or research design.
Version 1.03: Minor conceptual clarifications, consistency improvements, reviewer-informed refinements, and editorial updates across the programme.