IntegrateWise — Technical Glossary
SUPERSEDED (2026-08-31) — lineage only. Authority lives in the Founder Canon: docs/canonical/00-SPINEWORKSPACE-INTEGRATEWISE-CANON.md.
This document predates the 2026-08-31 canon and still uses retired terminology. Known drift: Twin Workbench → AI Workbench (ADR-2026-08-31-01); Twin Memory / Twin AI Memory → AI Continuity Memory (ADR-2026-08-31-02); the two-word form “Spine Workspace” is forbidden. Content left unedited as lineage. The OODA “control grammar” framing is retired — context actions are ambient per the canon.
Version: 1.0 | Date: 2026-08-23 | Status: ACTIVE Authority: Nirmal (Founder)Every technical term, product concept, and domain vocabulary used in the IntegrateWise / SpineWorkspace codebase.
Naming Hierarchy & Identity
| Term | Association |
|---|---|
| IntegrateWise | The company/brand. Maker of SpineWorkspace. |
| SpineWorkspace | The product. Domain spineworkspace.com. Self-organizing workspace where people and AI continue and complete work. |
| Workbench | The user’s single operational surface inside SpineWorkspace. Role-specific projection of work and context. |
| Customer Zero | IW’s own production tenant used to prove the package before external publication. |
| IW-Spine-Network | Locked public distribution identity for marketplace artifacts. |
| IW-Spine-Network Plugin | MCP-backed app with focused embedded UI for Plugin marketplace/directory. |
| IW-Spine-Network MCP | Remote MCP server exposing exactly 19 public tools for MCP marketplace/registry. |
The 12 Kernel Terms
| Term | Association |
|---|---|
| WORK | A human or machine moving an objective through state. |
| CONTEXT | The minimum relevant connected state required to understand current work. |
| SPINE | The canonical connected model of operational work state. System of record for entities, context, decisions, signals, proposals, approvals, outcomes. |
| CONTINUITY | Preservation and bounded transfer of relevant context across time, systems, humans, and agents. |
| WORKBENCH | The role-specific projection of work and context presented to a human. |
| TWIN | Non-authoritative cognitive participant that consumes grounded context. The user’s primary AI representative. |
| CAPABILITY | A registered operation that may read, derive, or change work state. |
| GOVERNANCE | The authority that determines whether a capability or proposed consequence is permitted. |
| HERMES | The execution coordinator for authorized capability invocations. |
| SYNC | Reconciliation of resulting state across connected systems. |
| BRIDGE | The governed exchange boundary through which an authorized consumer receives or contributes bounded continuity. |
| OUTCOME | The observed result of work or execution that may become new canonical state. |
OODA Grammar & The Four Buttons
| Term | Association |
|---|---|
| OODA | Observe-Orient-Decide-Act control grammar. The only OODA control grammar for truth capture, Twin reasoning, delegation, and governed execution. |
| Store in Spine | Observe button. User captures truth through governed Pipeline; Spine records entity, timeline, provenance. |
| Ask your Twin | Orient button. User asks Twin to understand the work; Twin reads bounded context then reasons. |
| Assign your Twin | Decide button. User delegates intent to Twin; Twin creates draft/proposal but does NOT execute until approved. |
| Approve Twin’s Action | Act button. User authorizes execution; Governance validates, Pipeline invokes, outcome reconciled to Spine. |
| Four Buttons | The explicit OODA control grammar — the only way to interact with Twin for consequential operations. |
Four Operational Patterns
The fundamental ways the workbench operates to enable human-AI collaboration. Not UI buttons — operational contracts.| Pattern | Association |
|---|---|
| Store in Spine | All business context, communications, decisions, and evidence are automatically captured into Spine Fabric. The organizational memory both humans and AI access. |
| Ask Your Twin | Human queries Twin through the workbench. Twin has access to full business context from Spine. Bidirectional — human can ask, Twin can proactively suggest. |
| Approve Your Twin | Twin’s proposals require human authorization for governed actions. Triggers: capability resolution → governance check → source system write → receipt → reconciliation. HITL by design. |
| Reject/Redo Your Twin | Human can reject Twin’s work. Twin retries with new context or escalates. System reconciles state. Ensures continuity, prevents hallucinations from cascading. |
Mapping to OODA
| Operational Pattern | OODA Phase | Button |
|---|---|---|
| Store in Spine | Observe | Store in Spine |
| Ask Your Twin | Orient | Ask your Twin |
| Approve Your Twin | Decide | Assign your Twin |
| Reject/Redo Your Twin | Act | Approve Twin’s Action |
Doctrines & Laws
| Term | Association |
|---|---|
| Twin Law | Twin is a capability of the Workbench, not a panel. |
| Context Law | The human never explains already-known work context to AI. |
| Memory Law | Spine owns truth. Memory informs truth. Memory is not truth. |
| Authority Law | Confidence increases trust. Confidence never grants authority. |
| Silence Law | Twin prefers silence over low-value commentary. |
| Capability Law | Workbench projects tool capabilities, does not duplicate tools. |
| Bridge Law | One platform. One boundary. Different projections. |
| Product-Platform Law | Broad underneath. Narrow at the surface. |
| Same-Artifact Law | Customer Zero and external customers use the same build. |
| One-Way Promotion Law | Promote only code/manifests; never tenant data/credentials/state. |
| Customer Zero Proof Law | CZ must run daily work through Plugin and browser Workbench. |
| Clean-Tenant Law | Frozen versions install into unrelated tenant without code changes. |
| Governed Execution Law | Every external action requires approval token, Hermes execution, audit, reconciliation. |
| Compatibility Law | Release manifest records compatible versions. |
| No Second Fortress | No product deployment becomes an independent authority plane. |
15 Canonical Spine Laws
| # | Law |
|---|---|
| 1 | No model owns truth |
| 2 | No agent bypasses Twin |
| 3 | No capability bypasses Governance |
| 4 | No frontend owns canonical state |
| 5 | No product owns credentials |
| 6 | No source-system mutation without receipt |
| 7 | No ambiguous retry |
| 8 | No context without authority |
| 9 | No observation without identity |
| 10 | No isolated agent memory |
| 11 | No second Fortress |
| 12 | No disconnected execution |
| 13 | No hidden incompleteness |
| 14 | No package authority |
| 15 | No provider-specific business UX |
14 Product Doctrines
Business need first, One View, One Twin, Agents are workers, Spine is contextual infrastructure, Context is shared, Handoffs are contextual, Capabilities are discoverable, Authority is explicit, Models do not own truth, Source systems remain execution authorities, Every meaningful action returns to Spine, Ask/Approve/Reject/Redo is native, More connected work strengthens Spine.Spine Fabric & Storage
| Term | Association |
|---|---|
| Spine Fabric | The organization’s persistent business-context substrate. Collection of storage and memory mechanisms, not one database. |
| Canonical Plane | Protects authoritative truth in Spine; governed writes only. |
| Operational Plane | Working/contextual projections + memory for Workspace, Twin, agents, workflows. |
| Canonical Truth | Authoritative business truth — entities, relationships, transactions, KPIs. |
| Operational Memory | Events, context, embeddings, semantic memory, active work state. |
| Adaptive Spine | The runtime realization of Spine with 7 canonical entity types. |
| Fortress | Legacy term for Supabase Postgres database used as canonical SSOT (historical/retired). |
| D1 | Cloudflare D1 — per-tenant operational Spine database; bounded edge state. |
| KV | Cloudflare KV — hot cache for rate limits, sessions, connector status, signal cache, metrics, OAuth state. |
| R2 | Cloudflare R2 — blob storage for files, attachments, evidence, raw payloads. |
| Vectorize | Cloudflare Vectorize — embedding storage for semantic search. |
| AI Search | Cloudflare AI Search — tenant-scoped semantic search over memory/knowledge. |
| Durable Objects | Cloudflare Durable Objects — agent memory with snapshotting to D1. |
| Neon Postgres | Canonical control-plane authority; owns canonical operational entities, relationships, lineage. |
| Six-Layer Storage Fabric | L1 Postgres/Neon/D1, L2 Redis/KV, L3 R2, L4 Conversational Store, L5 Vectorize/pgvector, L6 Durable Objects. |
Pipeline & Flows
| Term | Association |
|---|---|
| Pipeline | Data processing pipeline: LOAD → NORMALIZE → STORE → THINK → APPROVE → ACT → REPEAT. |
| Flow A | Structured operational data — deterministic normalization, governed writers, Pipeline promotion. |
| Flow B | Unstructured business data — documents, communications, attachments, transcripts. |
| Flow C | AI/agentic activity — never writes directly to Spine; Triage proposes promotion. |
| Flow D | Identity/authorization gate — tenant, workspace, role, configuration, connection, credential, capability, policy binding. |
| Normalizer | 6-stage data normalization kernel: NA0 Schema Detector → NA1 Canonical Transformer → NA2 SSOT Binder → NA3 Lineage Manager → NA4 Relation Binder → NA5 Spine Publisher. |
| Loader | Ingests raw source data into the pipeline. |
| Triage | Adaptive classifier/router for incoming information. Receives, validates, classifies, redacts, deduplicates, links, quarantines, stages. Cannot independently decide truth. |
| Memory Promoter | Engine that promotes selected Spine entities into Memory based on 7 Promotion Signals. |
| Promotion Signals | important, repeated, strategic, behavioral, decision, preference, exception. |
| Spine Writer | The governed sole-writer path to canonical state. |
| Deterministic Normalization | 8 stages: Raw Event Ingestion, Canonical Field Mapping, Schema Normalization, Entity Resolution, Deduplication, Relationship Linking, Depth Matrix Evaluation, Spine Write. |
| Depth Matrix | Tenant schema configuration controlling how deep normalization/relationship traversal goes. |
| Transactional Outbox | Domain write + outbox event happen atomically in D1; outbox publisher pushes to Cloudflare Queue. |
| Idempotency | Deduplication mechanism using processed_commands table; reused idempotency key with different payload = conflict. |
Capability Fabric
| Term | Association |
|---|---|
| Capability Fabric | Governed execution language between intelligence and operational action. Discovers, selects, validates, enforces governance, routes execution. |
| Semantic Capability | What a human or Twin means, expressed without any provider (e.g., email.send, crm.account.update). |
| Tool Capability | Versioned provider binding of a semantic capability (e.g., gmail.email.send@2). |
| ResolvedCapability | Fully bound triple: SemanticCapability + ToolCapability + ConnectionId. Nothing downstream may re-pick. |
| CapabilityRegistry | Registry of semantic capabilities and tool capabilities. |
| FabricCapabilityResolver | Resolves intent to resolved capability through registry + connections. |
| CapabilitySurfaceItem | Contextual, human-readable description of one capability for Workbench rendering. |
| MCP Pool | 55 registered capabilities. Pattern: Registry → Resolver → Health → Routing → Versioning. |
| Provider Management MCP | 15 capabilities: provider.*, apps.*, connections.*, oauth.*. |
| Integration Management MCP | 28 capabilities: connectors.*, sync.*, webhooks.*, schemas.*. |
| Task Management MCP | 12 capabilities: tasks.*. |
| EffectSignature | What one invocation of a tool capability changes in provider reality. |
| RiskClass | read, low, medium, high, critical — classification of capability risk. |
| Reversibility | reversible, compensatable, irreversible — whether an action can be undone. |
| Connection | Tenant-owned, credentialed instance of a connector definition. |
| ConnectionStatus | healthy, degraded, expired, revoked, unavailable. |
| EntitlementLevel | none, starter, growth, enterprise, unknown — access tier. |
Governance
| Term | Association |
|---|---|
| Governance | Policy and approval semantics — the operational control plane for whether actions may advance. |
| GovernanceDecision | Deterministic outcome: allowed, approval_required, or blocked. |
| PolicySnapshot | Policy state a decision was made against; snapshotting makes decisions reproducible and auditable. |
| HITL | Human-in-the-Loop. Governance behavior where operators act as reviewers, approvers, escalators, or exception handlers. |
| ApprovalState | draft, proposed, awaiting_human_review, approved, rejected, modified, deferred, executing, outcome_received, verification_required, verified, reconciled, closed. |
| AuthorityDecision | Record of a human decision on a proposal. |
| ProposalLifecycle | Full lifecycle of a proposed action from creation through verification. |
| Approval Token | Scoped human approval token minted by Governance; required for every external execution. |
Agents & Intelligence
| Term | Association |
|---|---|
| Twin Orchestrator | Cognitive orchestration layer. Observes, reasons, proposes, coordinates. The primary AI representative. |
| Three-Tier Agent Taxonomy | Primary Orchestrator (Twin), Cognitive/Governance Services (Triage, Governance, Entity Resolver, Evidence), Specialized Platform Workers (Schema Generator, Schema Orchestrator, Coworker Agent, Adaptive Auto-Tuner). |
| Triage Agent | Receives incoming information, validates, classifies, routes. Cannot decide truth independently. |
| Coworker Agent | Performs preparation and bounded delegated work. Proposes only. |
| Schema Generator | Proposes entities for schema-discovery coordination. |
| Schema Orchestrator | Maps/checks diffs in schema proposals. |
| Adaptive Auto-Tuner | Refines thresholds. Propose-only worker. |
| Entity Resolver | Cognitive service that resolves entity references during Orient/Decide. |
| Evidence | Bounded service Twin calls during Orient/Decide. |
| Agent Registry | Infrastructure service tracking available agents. |
| Tenant Brain DO | Durable Object for tenant-level governance and cognitive state. |
| Colony | Agent colony pattern for multi-agent runs. |
| Morning Brief | Scheduled 7am daily briefing from Twin. |
Identity & Auth
| Term | Association |
|---|---|
| One Auth | Single ecosystem entry for identity; host authority never duplicated. |
| One Access | MFE for identity, tenant/workspace context, sessions, consent, scoped grants, connector marketplace. |
| Tenant | Isolated customer environment. Every durable record carries tenant_id. Per-tenant D1 routing. |
| TenantGate | Gateway-enforced tenant boundary; every query binds tenant_id first. |
| Workspace | Organizational subdivision within a tenant; role-specific projection surface. |
| PrincipalContext | Identity and authority derived by Gateway from the One Access session. |
| ActorContext | Actor/authority context resolved by One Auth; never supplied by a browser. |
| Credential Wall | Where credentials and access tokens remain; never published to listing, browser, or MCP tool result. |
| Clerk | Primary authentication provider (JWT, OAuth). |
| Descope | Alternative auth provider for server-side session validation. |
| Cloudflare Access | MCP Auth provider. |
| RLS | Row-Level Security. PostgreSQL RLS enforcing tenant_id isolation. |
| WorkspaceRole | owner, admin, member, analyst, guest. |
| DomainId | integratewise-apac, personal, account-success, revops, salesops. |
Context & Memory
| Term | Association |
|---|---|
| ActiveContextBundle | One bounded, authorized context response for a work context. |
| Context Assembly | First-class system operation resolving relevant context for a business need. |
| Context Resolver | Assembles User, Work, Organization, Entity, Event, Document, Conversation, Transaction, Policy, and Capability context. |
| Context Bundle | Assembled context delivered to Human View and Twin Reasoning. |
| ContextScope | entity, relationships, knowledge, memory, evidence, capabilities, entity360, ai_output. |
| EntityContext | Disclosure available regardless of Entity 360 maturity. |
| Entity 360 | Single entity continuation view with activity, intelligence, action, governance, execution state, and continuity. |
| Entity 360 Projection | Optional, policy-gated projection requiring full Entity 360 maturity. |
| Entity 360 Formation | Process of building Entity 360 from resolved related entities, evidence, provenance. |
| FreshnessSummary | Data freshness: live, recent, stale, historical, mixed, unknown. |
| Memory Scopes | USER (per-person), WORK (per-entity/process), ORGANIZATION (tenant-wide). |
| Org Memory | Organizational memory stored in D1; promoted from Spine via Promotion Signals. |
| Continuity Bridge | Platform concept: operational reality → continuity capture → persisted state → context recovery → next actor context. |
| Workspace Discovery Projection | First browser read: GET /api/v1/workspace/discovery — versioned, tenant/RBAC-scoped manifest. |
| ContextGatewayAssertion | HMAC-signed claim asserting Gateway-resolved identity for context requests. |
| Canonical Memory Fabric | Platform-owned memory holding Working Memory and Agent Memory. |
| Working Memory | Short-term operational memory in Durable Objects. |
| Agent Memory | Long-term agent state in Durable Objects, snapshotting to D1. |
Projection & Workbench
| Term | Association |
|---|---|
| Projection Engine | Assembles workbench projections from Spine data. |
| ProjectionEvent | Event for projection snapshot, update, invalidation, execution outcome, notification, twin stream. |
| ProjectionChannel | projection.snapshot, projection.update, projection.invalidation, execution.outcome, notification.event, twin.stream, command, query, approval, presence. |
| WorkbenchCommand | Command from Workbench surface to platform. |
| WorkbenchShell | Top-level layout and context switching. |
| L1 | Internal engineering label for Workbench UI layer — never user-facing. |
| L2 | Internal label for Cognitive overlay (Twin). |
| L4 | Internal label for Continuity/Memory layer. |
| L7 | Internal label for Governance layer. |
| Five Zone Shell | Workbench layout structure. |
| Role-Based Workbench | Role-specific workbench rendering. |
| Department Workbench | Department-specific workbench. |
| OODA Buttons | UI component for the Four Buttons. |
| Twin Signal Feed | Inline signal display from Twin observations. |
| Store Spine Modal | UI for Store in Spine action. |
| Context Action Slot | Governed action component for context-aware actions. |
| Proposal Approval Card | UI for Twin proposal approval. |
Services
| Term | Association |
|---|---|
| Gateway | Single entry point for ALL requests. Validates JWT, resolves tenant, rate limits, routes to correct service. |
| Connector | Manages connections to external systems; sync, capability routes. |
| Connector Sync | Handles initial hydration and delta sync for connectors. |
| Pipeline | Data pipeline: normalize, spine write, memory promotion. |
| Intelligence | Think + Act + Govern + Agents cognitive layer. |
| Knowledge | Knowledge management service; search, embed, synthesize. |
| Hermes | Execution orchestrator for authorized capability invocations. |
| Store | File uploads, R2 storage, processing pipeline, version management. |
| Signals | Signal event bus with event sourcing; observe, replay, pattern matching. |
| Continuity | Persistent operational context service. |
| Twin Orchestrator | Cognitive orchestration: observe, reason, propose, coordinate, learn. |
| Loader | Raw data ingestion into the pipeline. |
| Normalizer | Data normalization service. |
| Govern | Governance service. |
| Act | Action execution service. |
| Think | Cognitive thinking service. |
| Admin | Admin operations service. |
| Billing | Billing management service. |
| Tenants | Tenant management service. |
| Webhook Ingress | Webhook ingestion and real-time broadcasting. |
| MCP Connector | Public 19-tool Spine MCP / semantic continuity boundary. |
| MCP Pool | MCP capability pool with 55 registered capabilities. |
| Telemetry | Application telemetry service. |
| Agent Registry | Agent registration and discovery. |
| IW Agent Runtime | Agent runtime with Durable Objects. |
| Task Management | Task CRUD, scheduling, cron, handlers. |
| Provider Management MCP | Provider, app, connection, OAuth management. |
| Integration Management MCP | Connector, sync, webhook, schema management. |
| Folder Watcher | Bridges local filesystem changes into pipeline. |
| HuggingFace Inference | HuggingFace model inference service. |
| Projection Service | Projection management service. |
Packages
| Term | Association |
|---|---|
@integratewise/kernel | Unified Kernel Connector connecting MCP with Spine, ADK, Orchestrator, Discovery. |
@integratewise/orchestrator | Execution engine managing capability execution across agents, connectors, MCP tools. |
@integratewise/capability | Canonical Capability Fabric runtime contracts. |
@integratewise/projection-contracts | Canonical types for Projection Engine and Workbench command bus. |
@integratewise/normalizer | 6-stage data normalization kernel. |
@integratewise/spine-entities | 7 canonical entity types for the Adaptive Spine. |
@integratewise/spine-network | IW Spine MCP Server connecting AI tools to IW ecosystem through the Spine. |
@integratewise/context-contracts | Active Context Bundle and Entity Context contracts. |
@integratewise/gateway-sdk | Gateway client SDK with spine-tools. |
@integratewise/model-gateway | Model registry and provider management. |
@integratewise/analytics | Analytics hooks and providers. |
@integratewise/webhooks | Webhook worker for incoming events. |
@integratewise/accelerators | Accelerator packs for different business domains. |
@integratewise/bootstrap | Bootstrap and initialization package. |
@integratewise/contracts | Shared contracts including authority doctrine, surfaces, wiring constants. |
@integratewise/handoff-adapters | Handoff adapters with idempotency and JSON adapters. |
@integratewise/workbench-config | Role configs, sidebar navigation, projection profiles. |
@integratewise/design-system | Design system components and utilities. |
@integratewise/skills | Skills definitions. |
@integratewise/types | Shared TypeScript types. |
@integratewise/config | Configuration including env bindings. |
@integratewise/core/governance-engine | Core governance engine with approval queue. |
@integratewise/coda-pack | Coda integration pack with sync tables and actions. |
Canonical Spine Entities
| Term | Association |
|---|---|
| Connection | Tenant-owned, credentialed instance of a connector. |
| Connector | Connector definition with capabilities. |
| SyncJob | Synchronization job tracking. |
| Webhook | Webhook registration and statistics. |
| Schema | Schema definition. |
| Capability | Capability definition with policy. |
| TimelineEvent | Temporal history of entity mutations; append-only. |
| PlatformEvent | Cross-service event with causation, correlation, idempotency. |
Execution & Reconciliation
| Term | Association |
|---|---|
| ExecutionPlan | Immutable plan created before Governance evaluation; carries content hash. |
| ExecutionStep | Steps: connector_call, agent_invoke, mcp_tool, spine_read, spine_write, approval_gate. |
| ExecutionLifecycleState | dispatched, provider_acknowledged, reconciling, reconciled, partially_reconciled, reconciliation_failed, promoted. |
| ReconciliationResult | Comparison of expected effects vs observed provider reality. |
| ObservedEffect | Effect actually observed from provider read-back or event. |
| EffectMatch | Comparison result for one expected effect. |
| PromotionEnvelope | Envelope consumed by Pipeline for canonical promotion; only promoted when reconciled. |
| ContinuityChain | Every identifier needed to reconstruct a continuity chain from one record. |
| RuntimeResult | Health/error result type for capability operations. |
| StructuredError | Typed error for capability operations. |
| Canonical JSON | Deterministic JSON serialization for content hashing. |
| PlanHash | SHA-256 over every plan field; re-derived downstream to detect tampering. |
| CapabilityIntent | User or Twin intent before any provider is chosen. |
| WorkService | Human operational work: tasks, queues, assignments, deadlines, ownership. |
| ExecutionService | Machine execution: provider dispatch, invocation attempts, retries, compensation. |
Projections & Workspace
| Term | Association |
|---|---|
| RoleConfig | Role definition: id, label, sidebar modules, connected systems, AI capabilities, operational features, MCP capabilities, Spine projection. |
| ModuleId | Sidebar module identifier. |
| PlatformCapability | Shared platform modules: integrations, configuration, activation, governance, continuity, network-graphs. |
| CTXEnum | CTX_CS, CTX_SALES, CTX_SUPPORT, CTX_PM, CTX_MARKETING, CTX_BIZOPS, CTX_TECH, CTX_HR, CTX_FINANCE, CTX_LEGAL. |
| 16 Canonical Sections | Home, My Work, Entity Database, Entity 360, Pipeline/Queue, Timeline, Calendar, Communications, Documents & Knowledge, Signals & Risks, Approvals, Planning Canvas, Reports & Metrics, Automation & Playbooks, Recent Work, Twin. |
| 15 Business Domain Workbenches | Account Success, Technical Account Success, Service, Support, Sales, Marketing, Content, Finance/RevOps, Product Engineering, Business Operations, Operations, Founder Ops, Human Resources, Legal, Personal. |
Four Planes
| Term | Association |
|---|---|
| Experience Plane | User Workbench, Twin, Admin, Mobile, AI Chat, Model Selector, Agent Canvas. |
| Intelligence Plane | Hermes, Knowledge, Memory, Signals, Model Gateway, Omni Router. |
| Execution Plane | Gateway, MCP Pool, Provider MCP, Integration MCP, Task MCP. |
| Foundation Plane | Adaptive Spine, Identity, Security, Storage, Timeline, Provenance. |
Two Capability Planes & Membranes
| Term | Association |
|---|---|
| Tool/Operational Capability Plane | ”What happened / what can execute”. Reads/mutates provider data; emits Action Receipts. |
| Cognitive Capability Plane | ”What might this mean”. Models/rerankers/embeddings/Twin; candidate proposals only (“rented cognition”). |
| Membrane A (Truth Egress) | Context Assembler redacts/restricts Spine context reaching cognitive models. |
| Membrane B (Truth Promotion) | Cognitive candidates pass Evidence Attachment → Triage → Governance → Human Approval before promotion. |
Data Concepts
| Term | Association |
|---|---|
| Provenance | Source and lineage of data; mandatory in universal record envelope. |
| Lineage | Source, source_id, event_id, timestamp trail. |
| Canonical Fields | Allowed fields per entity type for normalization. |
| Schema Normalization | Transforming raw data into canonical format. |
| Entity Resolution | Matching/merging entities across sources. |
| Deduplication | Removing duplicate records. |
| Relationship Linking | Connecting related entities. |
| Canonical Entity Types | account, contact, opportunity, deal, project, task, document, meeting, email, message, event, signal. |
| Universal Record Envelope | id, tenant_id, workspace_id, status, version, origin, confidence, source_connection_id, provenance, metadata, created_by/at, updated_by/at, archived_at — shared by every operational table. |
Marketplace & GTM
| Term | Association |
|---|---|
| Plugin Marketplace | External acquisition and embedded activation for AI hosts. |
| MCP Marketplace | External acquisition and tool-first activation. |
| Connector Marketplace | In-product capability activation and expansion. |
| Provider Marketplace | Provider-native IW packages when provider OAuth/webhook semantics require it. |
| MarketplaceInstallation | State machine: initiated → identity_required → workspace_required → consent_required → binding → hydrating → active. |
| Activation Hub | Onboarding completion and intelligence activation. |
| Integration Manager | Connectors, sync, webhooks, schemas management. |
| Configuration Manager | Tenant schema, depth matrix, policies. |
| Capability Manager | Maps connectors/tools into governed, permissioned capabilities. |
| North-Star Metric | Time to first approved governed outcome. |
| Claims Maturity Model | Defined / Implemented / Live-verified — code presence alone never satisfies a claim. |
Frontend Six Concepts
| Term | Association |
|---|---|
| Entity | Operational data entity. |
| Projection | Role-specific view of entity data. |
| Capability | Available operation. |
| Signal | Attention-requiring event. |
| Context | Relevant connected state. |
| Twin | Primary AI representative. |
Canonical Reference
For the complete operational model, see:WORKBENCH_INTERACTION_LIFECYCLE.mdx— The canonical Human-Twin-Agent flow, learning layer, and external tool integration
Tagline
“Truth you own. AI you rent. Approval in between.”