Skip to main content

Introduction to Watchlight AI

Watchlight AI Beacon is an enterprise platform for discovering, securing, and governing AI agent ecosystems. The platform uses a progressive deployment model — start with visibility, add authorization, then layer on runtime governance.

Three Tiers​

TierQuestionKey Capability
Discovery & RegistryWhat's running?Automatic discovery of AI agents and MCP servers with trust state management
AuthorizationWho can do what?Intent-based Cedar policies with goal management and delegation chains
Runtime GovernanceHow do we enforce it?Transparent proxy with policy enforcement and credential injection
QuickstartDeploy Discovery & Registry in 5 minutes
Design Partner SetupFull platform setup for design partners
Progressive DeploymentUnderstand the three-tier model
Python SDKIntegrate with your AI agents
API ReferenceComplete API documentation

Getting Started​

Choose your path based on your role:

For Developers​

  1. Quickstart — Deploy Tier 1 and see your agents in the registry
  2. Python SDK — Integrate authorization into your AI agents
  3. CrewAI / LangGraph / AutoGen — Framework integrations

For Control Plane Engineers​

  1. Architecture Overview — Understand how components fit together
  2. Progressive Deployment — Plan your rollout
  3. Enterprise Deployment — Production Kubernetes and Helm

For Security Teams​

  1. Authorization Concepts — Understand the Intent-Goal model
  2. Policy Examples — Real-world Cedar policy patterns
  3. Compliance — Regulatory framework alignment

Architecture Overview​

┌──────────────────────────────────────────────────────────────────────┐
│                         AI Agent Ecosystem                            │
│   LangGraph  │  CrewAI  │  AutoGen  │  Custom Agents                 │
└──────────────┬───────────────────────────────────────────────────────┘
               │
               ▼
┌──────────────────────────────────────────────────────────────────────┐
│  Tier 3: Runtime Governance                                           │
│  ┌─────────────────┐  ┌──────────────────────┐                       │
│  │    wl-proxy      │  │  wl-secrets-broker   │                       │
│  │  (policy eval +  │  │  (credential inject) │                       │
│  │   audit trail)   │  │                      │                       │
│  └────────┬─────────┘  └──────────────────────┘                       │
│           │                                                           │
│  Tier 2: Authorization                                                │
│  ┌────────▼─────────────────────────────────────────────────┐        │
│  │                      WL-APDP                              │        │
│  │  Intent → Goal → Delegation → Cedar Policy → Decision     │        │
│  └───────────────────────────────────────────────────────────┘        │
│                                                                       │
│  Tier 1: Discovery & Registry                                         │
│  ┌─────────────────┐         ┌───────────────────────────┐           │
│  │   wl-discover    │────────▶│       wl-registry         │           │
│  │   (scanner)      │         │   (service catalog)       │           │
│  └─────────────────┘         └───────────────────────────┘           │
└──────────────────────────────────────────────────────────────────────┘

Support​