This is Eng Labs. It is the part of the site where I write the way I actually think - short, opinionated, in the first person, and aimed at other people who build software for a living.
The blog is for the buyer. It answers questions a business owner or engineering lead asks before they hire us: what is an MCP server, what does an AI agent cost, how does Canadian data sovereignty work. It is polished, SEO-tuned, and written in the third person. That is the right tone for that audience, and it stays that way.
Labs is different. Labs is where I post field notes on things I am actually working on: a new model release and what it changes, an agent architecture I tried and whether it held up, a tool I shipped (or broke), a pattern I keep seeing across customer engagements. Less guide, more logbook.
What to expect
- Shorter than the blog. Most notes are 3-7 minutes. If a topic needs 2,000 words it probably belongs on the blog.
- Opinionated. I will tell you what I would actually do, not present five options and refuse to pick.
- First person. These are my notes. "I built", "I tried", "I was wrong about X."
- Diagrams and code. Real architecture sketches, not stock illustrations.
- Revisions. If I change my mind, I will update the note and say so.
What is here already
To start, two notes on agent infrastructure I ship in production:
- Why I built an autonomous task queue for Claude Code - the thesis behind Sugar, our open-source agent task queue.
- One guardrail file, every coding agent - the idea behind agent-guardrails.
How often
No schedule. I publish when I have something to say. If you want it when it lands, there is an RSS feed at /lab/feed.xml.
If a note is useful, disagree with it. The whole point is the conversation.