Content Intelligence
Content Intelligence is the governed path a piece of published content takes through the platform: it carries an author, a provenance record and a classification, it accrues interactions, it is ranked by cryptographic reputation rather than popularity, and it can be verified by anyone who receives it. Like every other Intelligence surface, it runs on the same cell, the same identity and classification model, and the same tamper-evident record as the rest of the platform.
The reachable slice: governed content, from author to reader
What is wired is the content lifecycle and the machinery around it, reachable through the workspace console's governed Feed tab:
- Reputation. A transparent ranker weights an author's provenance-derived trust as one visible ranking signal, reading the directory trust plane; it is fail-safe by construction, no score, a read failure, or an expired score all resolve to a neutral weight.
- Provenance. A public verification service resolves an author, a C2PA-style provenance manifest and event chain, a signature and an AI-involvement disclosure for a piece of content, and fails closed when the signer is absent.
- Interactions. Mentions, reshares, comments and notifications (behind a notification-preferences gate), the governed activity that accrues on content.
- Analysis and reach. A workspace-scoped governed feed with composition, a real-time (server-sent-events) stream tail, and cross-organization federation over ActivityPub (inbound delivery, outbound follow, actor discovery) so content can travel between estates.
- A fail-closed share pipeline. Every cross-boundary share runs an ordered gate chain, clean-room fan-out refusal, entitlement, body sealing for restricted posts, in which a missing or failing gate denies rather than leaks.
The content lifecycle
Figure: the content lifecycle. Content is published with a provenance record, ranked by cryptographic reputation (green), accrues governed interactions, and is verifiable by any reader; a cross-boundary share passes a fail-closed gate chain (amber, dashed) before it federates across organizations.
Table equivalent
| Stage | What it is | Scope note |
|---|---|---|
| Publish | Author writes content into a governed feed | Carries author, class and provenance |
| Reputation | Rank by provenance-derived cryptographic trust | Neutral on missing/stale data: never popularity |
| Interactions | Mentions, reshares, comments, notifications | Behind a notification-preferences gate |
| Share | Cross-boundary / cross-org share | Fail-closed gate chain; a missing gate denies |
| Federate | ActivityPub inbound + outbound | Content can travel between estates |
| Verify | Provenance manifest + signature + AI disclosure | Inbound verified; outbound signing not claimed |
Why this is a governed content surface, not a social feed
The difference is that every piece of content is a verifiable, classified, attributed object, not a row in an engagement database. A reader can check who authored it and whether its provenance chain holds; the ranking that decides what they see is a transparent cryptographic-trust signal rather than a hidden popularity model; and a share that would cross a workspace, a clean room or an organization boundary passes a gate chain that denies on any doubt. The governance travels with the content, which is exactly what an ungoverned feed cannot promise.
Across the deployment matrix
The lifecycle (publish, rank, interact, verify) runs on a single cell against its own directory and content stores, so it holds sovereign and air-gapped: reputation reads the local trust plane, and verification is a local check. It works single-cell, and where cells can reach each other the reputation reader simply misses on rows a cell does not yet hold and falls back to neutral, self-healing as the trust rows replicate. Cross-organization reach is the ActivityPub federation path; on an air-gapped cell that path is naturally quiet, and the governed local feed is unaffected.
Related topics
- Surveillance, the other Intelligence surface where inbound provenance is verified and outbound signing is not claimed.
- Discovery & Mapping, the estate-scanning surface that catalogues content at rest across a data estate.
- The catalog and lineage where governed content objects are catalogued alongside all other data.
- Governed reads, the read model every content access is subject to.
- The entity model, the principals an author and a trust score resolve against.