Concepts

• Organization and agent scope — Scope selects which organization and agent context reads and writes apply to.
• Agents and inheritance — Child agents inherit definitions; Studio shows the effective result.
• Sources and Memory — Sources feed Memory; Memory holds durable product records.
• Object type, content type, and data category — Identity, payload shape, and category are separate axes.
• Explicit metadata validation and installation — Validation does not install; installation requires confirmation.
• Effective metadata — The resolved definitions after inheritance and installation.
• Abstracts and bindings — An abstract is a virtual union projected through bindings — not stored rows.
• People union — People combines employees and contractors through explicit bindings.
• Mappings and transforms — Mappings describe how source shapes become Memory content.
• Raw records — Raw retains the original payload beside envelope metadata.
• Same-subject composition — Multiple content instances can belong to one subject when composition is explicit.
• Relationship definitions — A relationship definition is metadata describing allowed links — not a stored edge on a record.
• Stored relationship references — Records store relationship references that point at targets under a definition.
• Forward and inverse traversal — Traversal follows definitions forward or inverse, optionally expanding targets.
• Resolved and unresolved relationships — Unresolved targets are retained and shown — not silently dropped.
• Record revisions and stale-write conflicts — Writes carry an expected revision; stale writers cannot overwrite newer data.
• Analysis objects — Analysis captures evaluated evidence distinct from operational decisions.
• Decisions — Decisions record chosen actions distinct from analysis content.
• Intelligence proposals — Intelligence may propose metadata or actions but never auto-installs them.