Shane Deconinck

New Explainers: Agentic AI Infrastructure

I’ve been building interactive explainers on the protocols that make agentic AI work:

User OBO Agent Agent Gateway Cedar policies MCP Tools A2A Agents across organizational boundaries TSP Trust Spanning Protocol
  • OBO (OAuth On-Behalf-Of): How agents get delegated user authority
  • AgentGateway: Centralized policy enforcement
  • MCP (Model Context Protocol): How agents connect to tools
  • A2A (Agent-to-Agent): How agents talk to other agents
  • TSP (Trust Spanning Protocol): Trust across organizational boundaries

Each covers what the protocol solves, interactive step-through demos, security risks, and what’s still missing.