Work Normalisation & Systemic Clarity

Purpose

When people create tasks, instructions, and requests, they bring ambiguity, implied context, and unstated assumptions. Automation systems have no tolerance for these. The result is silent failures, misdirected work, and governance gaps that only surface during audits or incidents.

WNSC answers the question: what must be true about a work artifact before a system is permitted to act on it?

Scope

WNSC governs human-originated work artifacts. This includes:

• Tasks and to-dos created by or assigned to people
• Instructions and directives issued through automated channels
• Requests, tickets, and work orders originating from human intent
• Work items modified, escalated, or reassigned by people

WNSC explicitly does not govern:

• Datasets, telemetry, logs, or model inputs — those are governed by MASC
• Fully automated system-generated events with no human origin
• Read-only reporting and analytics outputs

Core Principles

Requirements

Normalisation

All work artifacts SHALL be transformed into a canonical structure before any system acts on them. Canonical fields include: domain, priority, stage, owner, intent, and constraint declarations. Aliases and informal variants must be mapped to canonical forms.

Explicit intent

Work artifacts MUST declare their intent, scope, and any relevant constraints. Implicit assumptions — "everyone knows what this means" — are not acceptable in a WNSC-compliant system.

Ambiguity handling

When a work artifact cannot be normalised unambiguously, it MUST be:

• Rejected with a clear explanation of what is missing
• Quarantined as a proposal pending human review
• Escalated to a human with the specific ambiguity identified

Guessing is not permitted. The system must surface its uncertainty explicitly.

Refusal rules

WNSC-compliant systems MUST be capable of refusing to act on work that violates:

• Safety constraints defined by the governing organisation
• Active compliance rules (GDPR, regulatory, contractual)
• Clarity thresholds — work that is too vague to execute safely

Refusal must be explicit and logged. Silent rejection is a WNSC violation.

Audit requirements

Every state change to a WNSC-governed work artifact must produce an immutable audit record that includes: the actor, the timestamp, the change made, and the rule or decision that authorised the change. This record must be queryable and exportable.

Ethical Safeguards

WNSC's person safety principle is operationalised through three specific prohibitions. These are not optional. They are structural requirements of any WNSC-compliant system.

These safeguards apply to all outputs of a WNSC-compliant system — including AI-generated summaries and narratives. A summary that implies a person is the cause of a bottleneck violates WNSC even if it does not name them directly.

Reference Implementation

WNSC is an abstract governance contract. Any system that satisfies its requirements may claim WNSC compliance.

Catalyst by Stratogenic AI is the reference implementation of WNSC. Every architectural decision in Catalyst reflects the WNSC principles:

• All work is normalised to canonical domains, priorities, and stages before entering the governance pipeline
• State mutations require explicit proposals — no silent updates to flow items
• Proposals are quarantined until accepted or declined by a human with authority
• Every acceptance and decline is ledger-recorded with actor, timestamp, and reasoning
• The AI narrative layer (Safe Narrative Engine) enforces WNSC person safety at the prompt level — it cannot evaluate individuals even if instructed to
• Ambiguous or conflicting items surface as conflicts in the GodThread, not as silently merged records

The /agents.md manifest that Catalyst exposes to agentic tools is an operational expression of WNSC — it instructs automated systems how to handle work safely and non-judgementally.

Relationship to MASC

WNSC and MASC (Minimum AI Standardisation Contract) are complementary frameworks that together define a complete governance layer for AI-assisted operations:

• WNSC governs the work itself — the human intent and instructions that initiate action
• MASC governs the data that work creates and consumes — ensuring AI systems act on trusted, traceable, quality-verified inputs

Neither framework is sufficient alone. WNSC without MASC produces well-governed work on untrustworthy data. MASC without WNSC produces trustworthy data from ungoverned work. Together, they ensure that the full chain — from human intent to AI action — is safe, auditable, and accountable.

Intellectual Property & Licensing

Related