Personal Role Layer

Two Kinds of Knowledge

Every person who holds a role accumulates two distinct kinds of knowledge. Personal knowledge — their skills, relationships, opinions, and individual insights — belongs to them. It travels with them when they move on and cannot ethically or practically be claimed by the organisation. Role knowledge — the decisions made, the processes built, the lessons learned, the workarounds discovered — belongs to the role. It should transfer to whoever holds that role next.

RolegacyAI is built on this distinction. The Personal Role Layer is the architectural mechanism that defines and enforces the boundary between these two domains.

What Stays with the Person

The personal layer includes:

• Individual professional skills and capabilities
• Personal network and relationships outside a professional role context
• Personal opinions, assessments, and perspectives
• Private communications and personal circumstances
• Performance history and individual development context

None of this belongs in the institutional memory. Attempting to capture and transfer it would be both ethically inappropriate and operationally counterproductive — a successor needs to know what the role has learned, not what the previous holder personally thought about their colleagues.

What Transfers with the Role

The role layer includes:

• Decisions made in the exercise of the role's responsibilities
• Lessons learned through the role's operational experience
• Workarounds discovered for broken processes, systems, or dependencies
• Vendor and partner relationship context developed in a professional capacity
• Process patterns and operational approaches built over time
• Technical configurations and institutional knowledge about systems
• Risk context, escalation paths, and operational guardrails

The Consent Model

RolegacyAI does not claim role knowledge without the involvement of the role holder. The capture and sanitisation pipeline surfaces memory candidates to the role holder for review and approval before they are committed to the institutional store. Role holders can reject candidates, edit them, depersonalise them, or mark them as personal. This consent model ensures the boundary is respected and that role holders are participants in the process rather than subjects of it.

Attribution and Anonymisation

By default, memory entries in RolegacyAI carry attribution — they are linked to the role holder who contributed them. This attribution supports trust (successors can see who documented what and when) and accuracy (reviewers can contextualise entries relative to when and by whom they were captured). Organisations can configure anonymisation policies where attribution is replaced with a role-level identifier, removing individual names from the institutional memory while preserving the temporal and contextual structure of the knowledge.