This record contains the manuscript “Procedural AI Assistants: A Governance-Aligned Architecture for Operational Decision Support.” The paper presents the design and governance model of a narrowly scoped artificial intelligence system intended to support operational staff in the consistent application of approved procedures. Unlike general-purpose conversational AI systems, the proposed architecture functions as a procedural assistant that provides guidance derived exclusively from a controlled knowledge base of standard operating procedures (SOPs). The system is intentionally designed with strict architectural constraints aimed at reducing governance, regulatory, and operational risks. These constraints include the absence of autonomous decision-making authority, the lack of execution capabilities, stateless processing without conversational memory, and the use of a closed procedural knowledge base rather than external information sources. User queries are processed through a hybrid retrieval pipeline combining deterministic routing, metadata-constrained filtering, and vector-based semantic search. In addition to describing the technical architecture of the system, the paper analyses how such constrained AI systems can align with contemporary AI governance frameworks. The discussion examines the system’s relationship with the European Union Artificial Intelligence Act, UK GDPR data protection principles, the NIST AI Risk Management Framework, and the OECD AI Principles. The paper introduces the concept of governance-by-design in the context of procedural AI systems, demonstrating how governance safeguards can be embedded directly into system architecture rather than implemented solely through organisational policies or external oversight mechanisms. The findings suggest that narrowly scoped AI systems designed with explicit architectural constraints may provide a practical pathway for deploying AI-assisted decision-support tools in operational environments while maintaining strong human oversight and proportional governance.