Run one agent
POST /api/agents/run with an agentId such as data_quality_agent or breach_analysis_agent.
Internal GCCAP operating surface
This page is retained behind the public client website.
GCCAP does not delete operating records, trial packs, hardware activation surfaces or evidence automation layers. They are preserved for authorised client portal, command-centre and internal operating use while the public website stays clean for airline first impressions.
Open portal accessInternal GCCAP surface
Run controlled AI recommendations across the GCCAP evidence chain.
The Agents Lab lets GCCAP run recommendation-only agents over current pilot data, then review agent output before any operational or client-facing action is taken.
StatusInternal / restricted surface. Not part of the public client walkthrough.
GCCAP
Sensor
Gateway
Evidence
Watchdog
Journey intelligenceRepresentative demo
Agents7 controlled roles
OutputRecommendations + audit
ApprovalHuman required
Operational layer
Agent lab capabilities
Use the admin APIs to run agents, review recommendations, update agent configuration, inspect audit records, and confirm production boundaries before any client-facing action.
Run all agents
POST /api/agents/run-all to generate a complete recommendation set from the current evidence chain.
Review recommendations
GET /api/agents/recommendations to inspect pending human-review items.
Approve or reject
POST /api/agents/recommendations/:id/review to record human review without silently altering source evidence.
Decision clarity
Build 10 admin API reference
The agent APIs are admin-gated and must not be exposed as public or client endpoints.
StatsGET /api/agents/stats
ConfigurationGET/POST /api/agents/config
RunPOST /api/agents/run
Run allPOST /api/agents/run-all
RunsGET /api/agents/runs
RecommendationsGET /api/agents/recommendations
AuditGET /api/agents/audit
Strongest next move
Do not confuse this with production AI automation
Build 10 is intentionally a controlled pilot-stage AI layer. Production AI requires provider selection, prompt/version controls, evaluation tests, guardrails, secure tools, cost limits, and formal governance.
Request a controlled pilot