ai-project-management project-artefacts key-decisions risk-registry pope-graph

What project artefacts should be in an AI-accessible memory?

Charter, goals, roles, decisions, risks, dependencies, status, retros. All POPE-tagged with owners and dates. bRRAIn's `Key-Decisions.md`, `Risk-Registry.md`, `Sessions.md`, and `NextSteps.md` templates operationalize this.

The eight artefacts every project needs

A project is agent-ready when eight artefacts are in memory: charter, goals, roles, decisions, risks, dependencies, status, and retros. Miss one and the agent starts guessing. Charter defines the boundary. Goals define success. Roles define who speaks for what. Decisions record irreversible choices. Risks flag what could break. Dependencies map the connective tissue. Status captures the current state. Retros feed lessons back in. bRRAIn's POPE graph treats each as a typed node with owners and dates, so the agent never confuses a proposal with a decision or a rumour with a risk.

Why POPE tagging matters

POPE stands for Person, Ontology, Process, Event — the four edges that make an artefact queryable rather than just stored. A decision with no person attached is a floating opinion. A risk with no date attached is a static worry. A dependency with no process context is a bullet on a wiki. bRRAIn's Consolidator enforces POPE tagging at ingest: every artefact enters the master context with owner, scope, and timestamp. Agents then answer "who owns this?" and "when was this last reviewed?" in a single graph lookup, not a Slack archaeology session.

The four canonical templates

bRRAIn ships four markdown templates that operationalize the eight artefacts. Key-Decisions.md captures each decision with its POPE fields and rationale. Risk-Registry.md tracks severity, likelihood, and owner per risk. Sessions.md records meeting context and outcomes. NextSteps.md tracks live commitments with due dates. These live at the root of every workspace and are edited like any other doc. The Handler reads them on every relevant query, so the PM updates one file and the entire team's agent experience updates with it.

What "agent-ready" unlocks

Once the eight artefacts are populated and POPE-tagged, the agent stops hallucinating at the first question about scope or ownership. A new engineer asking "who approved this architecture?" gets a name and a date. A stakeholder asking "what changed this sprint?" gets a diff. A risk manager asking "which open risks touch the Q3 launch?" gets a filtered list. The PM stops being the bottleneck because the artefacts the PM writes are now the interface the whole org reads from. Book a demo to see templates light up a project.

Relevant bRRAIn products and services

  • POPE Graph RAG — typed, queryable graph that makes each project artefact addressable with owner and date.
  • Consolidator — enforces POPE tagging at ingest and merges updated artefacts into the master context.
  • bRRAIn Workspaces — hosts the Key-Decisions.md, Risk-Registry.md, Sessions.md, and NextSteps.md templates.
  • Handler — reads the artefacts on every query so the PM's edits reach the whole team.
  • Ops Controller certification — teaches PMs to maintain the eight artefacts as live operational memory.

bRRAIn Team

Contributor at bRRAIn. Writing about institutional AI, knowledge management, and the future of work.

Enjoyed this post?

Subscribe for more insights on institutional AI.