| Term | Definition | Calls | Example |
|---|---|---|---|
| Primitive | Single, atomic, judgment-free operation. Does ONE thing. | Nothing | read_file, create_entity |
| Composite | Multiple primitives combined into a reusable operation. | Primitives | store_experience, execute_task_with_feedback |
| Workflow | Orchestrated sequence of composites with branching, conditions, and approvals. | Composites | meeting_notes_to_work_items, approval_workflow |
| Agent | Autonomous, goal-driven actor that selects and executes workflows/composites/primitives based on context. | Workflows, Composites, Primitives | meeting_assistant_agent, compliance_officer_agent |
| Orchestration | Meta-level coordination across multiple agents, handling conflicts, priorities, and system-wide coherence. | Agents | governance_orchestration, agent_fleet_orchestration |
| Term | Definition |
|---|---|
| Archer | Meta-orchestrator and UX layer. Persistent ambient presence that maintains system coherence. |
| ODIE | Outcome Driven Intelligence Engine. Reasoning engine for outcomes, beliefs, and signals. |
| Cogniscient | Adaptive intelligence substrate. Memory, knowledge graph, and experience storage. |
| Fluxio | Agent runtime and execution platform. Where agents run and actions happen. |
| Librarian | Resource manager. Scores and recommends tools, agents, and services. |
| Term | Definition |
|---|---|
| Outcome | A measurable desired state, structured as: Direction + Measure + Object + Context. |
| Signal | An observed data point that may indicate progress toward or away from an outcome. |
| Belief | A provisional assumption held with a confidence level, subject to revision based on evidence. |
| Entity | A node in the knowledge graph (person, organization, project, document, event, concept). |
| Relationship | A typed, weighted connection between two entities. |
| Experience | A stored summary of an interaction or event (gist, not verbatim). |
| Nudge | A non-invasive, actionable notification delivered in the user's flow of work. |
| Opportunity Score | (Importance × 2) - Satisfaction. Measures how underserved an outcome is. |
| Term | Definition |
|---|---|
| Parity | Whatever the user can do via UI, the agent can do via tools. |
| Granularity | Prefer atomic primitives. Features emerge from agent loops, not bundled tools. |
| Composability | New features = new prompts (when tools are atomic and parity exists). |
| Emergent Capability | Agent can accomplish things you didn't explicitly design for. |
| Judgment-Free | Tools don't decide IF or WHICH. Judgment belongs in prompts. |
| Term | Definition |
|---|---|
| Autonomy Level | How much authority an agent has (0=observe only → 5=broad autonomy). |
| Approval | Human authorization required before an action executes. |
| Policy | A rule that governs what actions are allowed, forbidden, or require approval. |
| Audit Log | Immutable record of all actions taken, by whom, when, and with what result. |
Document generated January 2026