Layer 2 Artefacts: Governance in Use
Layer 2 artefacts exist while the agent is operating in real workflows. They ensure that trust is earned through evidence, not assumed, and that autonomy changes are deliberate,
documented, and reversible. These artefacts make governance visible without creating friction.
6. Trust & Validation Log
The Trust & Validation Log exists to capture evidence that supports or challenges trust in the agent during real use. It is created once the agent is operating and is updated
continuously. This artefact is the practical foundation of “trust-but-verify” because it turns trust into something observable, discussable, and evidence-based rather than opinion.
The log captures adoption signals, validation outcomes, corrections or overrides, and observed failure patterns. Its value is that it provides an ongoing record of performance and
reliability, allowing teams to make confident decisions about autonomy and scope without relying on gut feel.
Download template →
7. Autonomy Progression Record
The Autonomy Progression Record exists to document any change in autonomy. It is created whenever autonomy is increased or reduced because autonomy is treated as a managed variable,
not a feature. Without this record, autonomy changes become informal, hidden, and difficult to reverse.
This artefact captures the previous autonomy level, the new level, the evidence supporting the change, the risks considered, and the reversibility plan. Its purpose is to ensure
autonomy changes are deliberate, justified by validation, and always accompanied by a clear way back.
Download template →
8. Risk Context Assessment
The Risk Context Assessment exists to assess how contextual changes affect risk. It is created when scope changes, new data is introduced, or the agent is reused in a new environment.
This is essential because risk does not live inside the agent alone; it emerges from the interaction between the agent, the task, the data, and the environment.
The assessment captures contextual risk factors such as regulatory constraints, reputational exposure, data sensitivity, and operational dependence, and records the mitigation approach.
It also clarifies how the risk profile impacts autonomy and controls. This artefact prevents organisations from mistakenly assuming that an agent that was safe in one context remains safe
in another.
Download template →
9. Reversibility Plan
The Reversibility Plan exists to define how autonomy can be safely reduced or paused. It is created before autonomy is increased because reversibility is the mechanism that makes progress
sustainable. If autonomy cannot be reduced cleanly, it should not be increased.
The plan defines trigger conditions for rollback, the responsible decision-maker, and the operational steps required to reduce autonomy. It reinforces that rollback is not a failure mechanism.
It is an engineering feature of responsible governance that preserves confidence as capability expands.
Download template →
10. Governance Decision Log
The Governance Decision Log exists to record key governance decisions without centralising power. It is created whenever a material governance decision is made, ensuring visibility and traceability
while keeping decision-making close to delivery.
It records the decision taken, evidence considered, roles involved, and the review point. Its purpose is not bureaucracy. It is to preserve institutional memory, reduce re-litigation of decisions,
and maintain confidence that governance is deliberate rather than ad hoc.
Download template →
