The parts of an AI workflow I deliberately keep boring
Where I use AI, where I use ordinary code, and why the handoff between them matters. A practical look at document automation, with a worked example and a downloadable test pack.
Insights
Grouped by what you're trying to do, not when it shipped — operators, builders, and marketers staying visible.
For operators
How to pick AI projects that actually pay back, and what it takes to get them past the proof-of-concept graveyard into production.
Sep 8, 2026 · 8 min read
Where I use AI, where I use ordinary code, and why the handoff between them matters. A practical look at document automation, with a worked example and a downloadable test pack.
Apr 8, 2026 · 12 min read
A first-person story about growing up an entrepreneur, always building things, always solving problems, and the moment AI arrived and quietly turned every idea I'd been carrying for twenty years into something I could actually ship.
For builders
The fluency layer. Prompting, context engineering, and the component vocabulary that makes AI-assisted building actually work.
Apr 19, 2026 · 10 min read
Most teams trying to improve their AI output are tweaking prompt wording. That is the wrong lever. Here is why context, not prompt craft, is the thing that actually moves quality, and how to build a context library that does it right.
Apr 21, 2026 · 11 min read
Every UI component you ship with AI tools answers to two audiences: the model you're using to build it, and the AI crawler that has to read it. Most teams work with neither in mind. Here's the vocabulary that makes both easier.
Browse everything
The full archive, newest first. Filter by tag to narrow in — curated clusters above cover the work that matters most.
Where I use AI, where I use ordinary code, and why the handoff between them matters. A practical look at document automation, with a worked example and a downloadable test pack.
Every UI component you ship with AI tools answers to two audiences: the model you're using to build it, and the AI crawler that has to read it. Most teams work with neither in mind. Here's the vocabulary that makes both easier.
Most teams trying to improve their AI output are tweaking prompt wording. That is the wrong lever. Here is why context, not prompt craft, is the thing that actually moves quality, and how to build a context library that does it right.
A first-person story about growing up an entrepreneur, always building things, always solving problems, and the moment AI arrived and quietly turned every idea I'd been carrying for twenty years into something I could actually ship.
Perspectives on AI implementation, automation strategy, and marketing engineering. Delivered when I have something worth saying.