AI Tools
Less baggage. More room to think.
Meet Context Diet: a Codex plugin that uses Jev or local Laya to sort useful context from clutter. Why I built it, how it works, and how to try it.
7 min read
The journal
Blog
Field notes on AI strategy, engineering leadership, platform architecture, and shipping practical systems.
Start here
AI strategy How to separate prediction, generation, and business value. AI infrastructure Architecture choices for retrieval, context, and model workflows. Leadership Current role notes and engineering leadership context. Developer workflow Practical field reports from local models and agent tooling.Latest notes
Chronological archive, newest first.
22 published posts
7 min read
AI Strategy
The Model Is Only Half the Agent: What an Agentic Harness Actually Does
Why the same AI model feels different across products, and how tools, context, memory, agent loops, guardrails, and verification turn it into an agent.
8 min read
AI Leadership
AI Is Changing the Shape of Work
AI will automate more routine execution, but the bigger shift is how roles are shaped. The advantage will go to teams that move people toward judgment, ownership, and AI supervision.
7 min read
AI Strategy
Predictive AI vs Generative AI: What Each One Is Actually For
Predictive AI forecasts measurable outcomes; generative AI creates new content. Here's how they differ, where LLMs fit, and how teams should combine them.
3 min read
Personal Notes
Stepping Into a New Role at WP Engine
I’ve stepped into a new role as Senior Engineering Manager at WP Engine, where I’ll be building and leading the engineering team focused on AI.
7 min read
AI Infrastructure
Is RAG Still Relevant in 2026? Choosing Between RAG and Long Context for LLM Apps
RAG is not dead, but it is no longer the default answer to every LLM architecture problem. This post breaks down where long context wins, where retrieval still matters, and how to choose the right approach for modern AI systems.
8 min read
Developer Workflow
Run Qwen 3.5 35B-A3B Locally as a Claude Code Assistant
A practical step-by-step guide to running unsloth/qwen3.5-35b-a3b in LM Studio and wiring Claude Code to localhost:1234 for a private local AI coding workflow.
11 min read
AI Infrastructure
Supercharge Your AI Workflows: GLM-4.7 + Cloudflare AI Gateway Integration
A step-by-step guide to integrating Z.AI's GLM-4.7 model with Cloudflare's AI Gateway as a custom provider. Learn how to configure caching, rate limiting, and authentication for optimized AI API workflows with real-world examples and complete cURL request configurations.
2 min read
Personal Notes
2025, A Year of Doing the Right Thing
Reflecting on a year defined by integrity, curiosity, and movement. From receiving WP Engine's Supercharged Award to making 3800 experimental git commits, exploring AI as a Z.AI ambassador, and traveling twenty thousand miles across continents. 2025 was about staying honest while building work that matters.
8 min read
Developer Workflow
Building Faster With MCP, GLM 4.6 and Claude Code CLI, A Field Report
A practical field report on using GLM 4.6, MCP servers, and Claude Code CLI to build a landing page. This post breaks down how AI-powered agents with live documentation retrieval and workspace awareness can genuinely accelerate engineering workflows—reducing context switching and delivering real leverage instead of just automation.