Artefacts
What are artefacts
Gleo's long memory — the structured outputs and references that outlive any single session.
Sessions are bursts. Decisions get made, ideas get captured, deliverables get sketched. The next week, the next playbook, the next quarter — most of that disappears unless someone made a real effort to file it somewhere durable.
Artefacts are Gleo's answer to that. They're the structured outputs and references that outlive any single session — the decisions, the strategies, the deliverables, the documents your team produced or pointed to. They sit at the company or playbook level, get versioned over time, and stay searchable long after the session that created them is in the rear-view.
The four dimensions
Every artefact in Gleo is described by four things. Once you know the four, the rest of this category falls into place.
Kind
Where the artefact came from and how it's edited: managed, derived, generated, or referenced.
Category
What the artefact is about: a decision, an action, a question, a risk, a deliverable, strategy, policy, or vision.
Content mode
Where the body lives: embedded (in Gleo), external (a pointer), or snapshot (a point-in-time copy).
Status
Where the artefact is in its lifecycle: draft, proposed, confirmed, active, superseded, or archived.
The dimensions are independent — pick a value for each. A single artefact might be a managed decision that's embedded and active. Another might be a referenced deliverable that's external and proposed.
Kind, in detail
The four kinds describe how the artefact is created and maintained:
| Kind | What it means | Typical example |
|---|---|---|
| Managed | Owned by a human. Edited by hand; promoted through statuses deliberately. | A team's working strategy document, a policy, a vision statement. |
| Derived | Produced from other artefacts or session outputs as a transformation. Re-derivable if its inputs change. | A roll-up summary of decisions across all sessions in a playbook. |
| Generated | AI-generated from session content. Re-generatable on demand. | A synthesised retro from prep responses; a draft executive summary. |
| Referenced | A pointer to something that already exists elsewhere — Gleo doesn't author the body. | A link to the team's actual product roadmap in Notion. |
"Generated" vs "Managed" is the most important distinction. Generated artefacts are AI output and can be re-generated at will — treat them as proposals, not source-of-truth. Managed artefacts are owned by humans and changes need to be deliberate.
Category, in detail
Category is the what — a coarse semantic tag that makes artefacts findable and lets the team see at a glance what kind of thinking they're looking at:
- Decisions — calls the team has made and stand by.
- Actions — work to be done.
- Questions — open questions, not yet decided.
- Risks — things to watch out for.
- Deliverables — outputs the team is producing.
- Strategy — the team's thinking on direction.
- Policy — rules the team operates by.
- Vision — where the team is headed at a longer time-horizon.
Categories are deliberately small and durable; the full filter set appears in the Artefact Library.
Content mode, in detail
Content mode is the where the body lives dimension:
- Embedded — Gleo holds the content. You edit it inside Gleo's editor.
- External — A pointer to a source-of-truth that lives in another system (Google Docs, Notion, Figma, Miro, Confluence). Gleo doesn't own the body. See External artefacts.
- Snapshot — A point-in-time copy taken from an external source. The original keeps changing; the snapshot doesn't.
Pick the mode that matches reality for that artefact. If the team is going to keep editing it in Gleo, embed it. If the source-of-truth lives in Notion, reference it externally. If you need to capture "what it looked like at the point we shipped the proposal," snapshot it.
Status, in detail
Status is the lifecycle position — covered in depth in Statuses and versioning, but the short version:
draft → proposed → confirmed → active → superseded → archived
Most artefacts spend most of their life in active. The transitions in and out are deliberate human decisions; status is not a label you set casually.
How the dimensions play together
A worked example. Suppose a team has been running a strategic-planning playbook for a client. After the third session, they have:
- A managed, strategy artefact that's embedded and active — the agreed three-year plan.
- A generated, decisions artefact that's embedded and proposed — an AI-synthesised list of decisions from the last three sessions, awaiting confirmation.
- A referenced, deliverables artefact that's external and active — a pointer to the client's actual product roadmap in Notion.
- A managed, risks artefact that's embedded and draft — a partially-written risk log, not yet circulated.
All four show up in the same Artefact Library, filterable along any of the four dimensions.
Tips & pitfalls
Don't over-classify on day one. Pick a kind, pick a category, save it. Refining the rest is cheap; getting started matters more.
Generated artefacts are starting points, not endings. When the AI produces a synthesis, treat it as a proposed artefact a human still needs to confirm. The lifecycle is built around exactly this kind of handover.