Connector System Overview

Connectors integrate external systems into Adspire.AI. They are declarative: you describe what the connector does (events in/out, actions exposed), and the platform handles the delivery and orchestration.

Core principles

• Connector = declarative integration All connectors should describe what they do (events in/out, actions exposed), and the platform handles delivery and orchestration.
• Event-first architecture All signals and UI actions should map to events. Transport details are abstracted by the platform.
• Category-based UX UI should present connectors by category (UI Extensions, Data Signals, System), but allow filtering by capability.
• AI-native context layer Connectors should expose structured context for agents, not just events and UI actions.

Connector scope

1. Workspace-specific Purpose: extend the Adspire interface with actions or mini workflows.
2. Brand-specific Purpose: extend the Brands interface with actions or mini workflows.

Connector categories

1. Data Purpose: ingest events into the Brand Graph and signal layer.
2. Actions Purpose: extend the Adspire interface with actions or mini workflows.
3. System Purpose: infrastructure-level integrations.

Connector capabilities

Common capabilities include:

ingest, publish, retrieve, search, transform, sync, notify, generate, analyze, approve

Connector definition

Add these fields to the connector manifest:

• scope: workspace | brand | org
• surfaces: actions, signals, resources, ai_context
• capabilities: ingest, publish, retrieve
• entities: product, blog, asset, campaign
• permissions: read_products, write_posts, read_assets
• ai_context: what the connector contributes to agents
• ui_contexts: where actions appear
• event_contracts: declared input/output event schemas

Connector model

Axis	Values	Why it helps
Type	workspace, brand, org, personal	Clarifies scope and permission boundary
Surface	action, signal, resource, system, ai context	Separates UI actions from data ingestion and backend utilities
Capability	publish, ingest, sync, search, generate, analyze, approve, store	Lets the UI and agents filter by what it can do
Direction	inbound, outbound, bidirectional	Makes sync semantics explicit
Runtime	webhook, polling, API, event stream	Keeps implementation flexible
Trust level	read-only, write, admin, restricted	Important for AI agent safety

Signal definition

For data signals, explicitly define:

• source_object
• event_name
• dedupe_key
• sync_mode: full | incremental | webhook
• freshness_sla
• mapping_rules
• entity_resolution