Strategy
Your org chart is a fossil of old information costs
AI isn't automating your tasks β it's repricing coordination. From Coase and Taylor to two landmark 2023 field studies: how to re-derive your company's layers.
Blog
Insights from the operating layer.
How enterprise AI actually gets deployed β from the team building the rails.
Strategy
The A2A protocol, explained: how AI agents learn to work together
Agent2Agent (A2A) is the open standard that lets AI agents from different vendors discover each other and delegate work. Here is how it works, how it differs from MCP, and why interoperability is a business decision.
Engineering
What MCP really is β the shipping container for AI agents
What the Model Context Protocol (MCP) is, how it differs from tool calling and A2A, and why standardizing how agents connect to your systems still leaves identity, approvals, and cost as your problem to solve.
Reliability
Why traces matter: you cannot operate AI you cannot see
A trace is the flight recorder of an AI workflow. Here is why tracing is the difference between debugging in minutes and guessing for days, with the failure cases that happen when you fly blind.
Governance
Your compliance team is the fastest path to AI production β not the roadblock
In regulated enterprises, the legal review isn't what stalls AI. The absence of an audit trail is. Here's how to turn governance into a deployment accelerator.
Product
Start with the experience, then work backwards to the model
The design tradition that reshaped consumer technology insists you begin with the human experience and work back to the technology β never the reverse. Enterprise AI almost always does it backwards.