Blog
Writing on AI engineering, agentic systems, and Context Engineering — what I’m building, what’s breaking, and what I’m learning.
AI coding sessions have a cold-start problem (the agent walks in blind) and a cold-stop problem (the next session inherits nothing). Two boring slash commands — `/primer` at the start, `/handoff` at the end — fix both. Here's what I run, why it works, and the honest tradeoff.
AI coding feels magical for a few prompts, then turns into chaos. The fix isn't a better tool — it's a tighter workflow. Here's the per-ticket loop and the larger flow around it that I use to ship real software with agents instead of just demos.
70–85% of enterprise AI projects never reach production. 95% of generative-AI initiatives create no measurable business impact. The companies winning aren't the ones who picked the "right" model — they're the ones who solved the unglamorous problems first.
Most consumer AI adoption is still surface-level — better questions to a chat box. The real lift comes from connecting AI to the data, systems, workflows, and decisions where work actually happens. Here's why that gap matters and what comes next.
Thoughts on AI engineering, agentic systems, and Context Engineering — plus the unglamorous production work that separates demos from systems. Here's what you'll find on this site, what I plan to write about, and how to reach me.
