Field guides

Deep, reference-grade documents written after building the thing. Protocols before they are mainstream, methods proven on production workloads, honest about what is simulated and what would break. Bookmark tier or nothing.

AG-UI Deep dive · 18 min

AG-UI: Streaming the Interface, Not Just the Answer

A field guide to the agent-user interaction protocol, and what happens when a process engine drives the stream

A field guide to AG-UI, the agent-to-user event protocol: event anatomy, transports, process-orchestrated generative UI, and what it leaves unsolved.

AP2 v0.2 Deep dive · 16 min

AP2: The Agent Payments Protocol

What I learned making an agent that buys FIFA tickets while I sleep

A builder's deep dive into AP2 v0.2: mandate cryptography, the four-party model, threat model, and Human-Not-Present flows, verified against the spec.

Deep dive · 22 min

Where Agents Actually Work: Field Notes from 50+ Enterprise Builds

Two years, seven industries, four workflow shapes. A map of where agentic AI earns its keep, and where plain code still wins.

Field notes from 50+ enterprise agentic AI builds: where agents earn their keep, by industry and workflow shape, and where plain code still wins.

Deep dive · 22 min

The Control Plane: Governing Agentic AI at Scale

Agent frameworks own the inner loop. Someone has to own the outer one.

Agent frameworks own the inner loop. Your institution owns the outer one. A framework-agnostic pattern for governing agentic AI in production.

Deep dive · 20 min

Evaluating Agentic Systems: The Four-Axis Method

Cost per case, per-field quality, tool-skip rates, and temperature: how to pick a model for an agent that lives inside a governed process

A field-tested method for evaluating process-embedded AI agents: cost per case, per-field accuracy, tool-skip rates, and temperature sensitivity.

MCP Deep dive · 17 min

Tool Contracts: Why Agents Skip Your Tools

A tool spec is a contract between a deterministic system and a probabilistic caller. Most teams write it for the wrong reader.

Why agents silently skip tools: a failure taxonomy, the spec rules that measurably cut skip rates, and a deterministic 0-100 linter for MCP tool contracts.

A2A Deep dive · 22 min

A2A: Agent-to-Agent, a Field Guide for Builders

How agents from different vendors, teams, and trust zones actually talk to each other, and how to keep it auditable

A builder's field guide to the A2A protocol: Agent Cards, task lifecycle, orchestration vs swarm, privacy boundaries, and the threat model that matters.

> esc