Blog / Agent interoperability, in depth
Writing on the durable work object
Field notes on AI agent buses, the intent envelope, human-in-the-loop execution, and coordinating agents with workflow tools — without losing context at the handoffs.
MCP Certification Arrives: What the New MCPA Exam Signals for MCP Security and Governance
The Agentic AI Foundation launched the MCPA, the first official MCP certification: a 120-minute exam where Interactions & Execution (26%) and Security & Governance (24%) make up half the score. What that weighting means for teams running agents.
Read article →MCP vs A2A in Fall 2026: What Each Protocol Standardizes and the Execution Gap Between Them
MCP standardizes how one agent reaches tools and data. A2A standardizes how independent agents delegate tasks to each other. Both now sit under the Agentic AI Foundation. What each spec covers, where they meet, and the execution layer both leave to you.
Read article →How to Orchestrate AI Agents and n8n Workflows Without Losing Context
n8n is great at wiring actions; agents are great at deciding. The gap is a durable object that survives the handoff between them. Here is how to close it.
Read article →Human-in-the-Loop Agent Execution as a Native State, Not an Afterthought
Most agent stacks bolt human approval on as an afterthought. Treating review as a native run state is what makes mixed execution actually usable.
Read article →The Intent Envelope, Explained: A Durable Object for Agent Work
A prompt is not portable. An intent envelope is. A field-by-field walk through the durable object at the center of the Agintent protocol.
Read article →What Is an AI Agent Bus? The Coordination Layer Agent Stacks Are Missing
Most agent stacks fail not at generation but at the handoffs. An AI agent bus fixes the missing coordination layer with one durable work object.
Read article →