Memory Map, expertise and the knowledge library

Once claims are approved they are readable three ways, and which one you want depends on the shape of your question.

Knowledge assets

“What do we have written down?” The library, by lifecycle state and type.

Memory Map

“How do these things relate?” Projects, people and systems as a graph of verified relationships.

Expertise

“Who actually knows this?” Candidates ranked from evidence rather than from a directory.

Knowledge assets is verified, versioned memory. Search it, filter it, and open any asset to inspect its claims, its owners, and where each claim came from.

Lifecycle states

Types

An asset is one of: process, decision, runbook, faq, system, risk, stakeholder, dependency. The type is not decoration. Risk scoring and the continuity pack both read it, and a decision record with no rationale is treated differently from a runbook with no steps.

1. 1

   Filter to what you are auditing

   Status and type filters combine. Published + runbook answers “what operational procedure have we actually certified?”

2. 2

   Open an asset and read its claims

   Each claim shows the evidence it is quoted from. This is the screen to open when somebody asks “where did that come from?”

3. 3

   Check its owners

   Ownership is what drives continuity risk. An asset whose only owner is one person is exactly what chapter 11 is about.

Memory Map holds projects, people, systems and decisions, and the verified edges between them. It is populated from the Relationships lane of the review queue: an edge exists here because a human approved it, and for no other reason.

1. 1

   Find a project

   The list shows every project memory that has reached the workspace, with a count of how many edges on it are still to review.

2. 2

   Open it for the dependency graph

   People, systems, decisions and the links between them, on a full canvas.

3. 3

   Follow an edge to its claim

   Any relationship opens onto the claim behind it: its evidence and its blast radius, meaning what else in the graph depends on this edge being true.

Expertise answers who knows this? by searching verified ownership, responsibilities and evidence. It ranks candidates from what they have demonstrably touched, not from a job title, a team page or a self-declared skill list.

1. 1

   Describe the area, not the person

   “Atlas release process”, “payment reconciliation”, “the incident escalation path”.

2. 2

   Press Resolve

   You get ranked candidates, each with the evidence that put them there.

3. 3

   Read the evidence before you route work

   A candidate ranked on one stale document is a different answer from one ranked on six current ones, and only the evidence tells you which you have.

• Use it before a handover to find who could take something on.
• Use it during an incident to find who has actually touched the system, rather than who is on a rota.
• Use it when Ask abstains. The person may know what memory does not, and that gap is what a capture request is for.