Engineering journal / 2026

Notes from
inside the system.

Writing on agent architecture, software systems, and the engineering decisions that make intelligence useful.

12 field notes
Latest Field NotesFeatured Analysis
All writingTopic / reading time
04

Agentic Architecture / 24 min

Standardized Interoperability Is Connecting Every Digital FTE.

How open protocols, Model Context Protocol (MCP), SKILL.md procedures, and Agent-to-Agent (A2A) message buses unite isolated AI workers into scalable, enterprise-grade autonomous swarms.

05

Enterprise AI / 22 min

Enterprise Agents Are Building The Workforce.

From tools for tasks to a system of record. The definitive 2026 technical analysis of how enterprise AI agents transition from stateless task tools into durable, persistent systems of record.

06

AI / 25 min

The Vertical AI Employee: Building the Future of Domain-Specific Autonomous Agents

The definitive 2026 blueprint for building persistent, secure, and domain-specialized AI employees. A deep dive into the Body+Brain architecture, Agent Skills, and MCP standards.

07

Engineering / 20 min

Agents Building Agents: The Recursive Power of Claude Code

Exploring the shift from manual coding to meta-tooling. How Claude Code enables a recursive development loop where AI agents autonomously build, secure, and extend enterprise-grade Vertical AI Employees.

08

Agent Architecture / 18 min

Designing Systems Around Models, Not Demos: The AI Agent Factory Blueprint for Production AI Workers

The Third Era of AI is not about better chatbots—it's about Digital FTEs. Learn how the Agent Factory's spec-driven methodology, Four Layers architecture, and Job-to-be-Done framework transform models into reliable AI Workers with verified outcomes.

09

Harness Engineering / 20 min

The Agent Harness: Agent Factory's Blueprint for the Runtime that Makes AI Workers Controllable and Useful

The model is 10% of an AI Worker. The Agent Factory framework's harness engineering—covering MCP sandboxes, loop engineering, state checkpointing, OpenClaw, and Hermes—is the 90% that makes agents reliable in production.

10

Context Engineering / 17 min

Context Engineering: The Agent Factory's System of Context, SKILL.md Pattern, and AI-Searchable Knowledge Architecture

The AI Agent Factory distinguishes between Systems of Record (where facts live) and Systems of Context (how an AI Worker reasons about them). Learn to build the five-layer context architecture with Neon Postgres, pgvector, and SKILL.md procedures.

11

Deterministic AI / 19 min

Deterministic AI: The Agent Factory's Spec-Driven Development, State Machines, and require_approval Architecture

The AI Agent Factory's Spec-Driven Development methodology pairs non-deterministic LLM reasoning with deterministic state machines, typed output contracts, idempotent tool calls, and the require_approval human gate for zero-hallucination enterprise workflows.

12

AI Reliability / 22 min

AI Reliability Engineering: The Agent Factory's Eval-Driven Development, Maker-Checker Pattern, and 'Leaving the Laptop' Standard

The AI Agent Factory's Eval-Driven Development methodology, Maker-Checker pattern, distributed tracing, and the 'Leaving the Laptop' standard define what it means to build AI Workers that humans can genuinely trust.