Field Notes · Building with AI
What I'm building,
and why.
I'm a marketer who builds his own systems. This is where I write down what I'm making with AI — the reason for it, how I set it up, and the tools I reached for. Working notes, not tutorials.
Why I bought 128GB of RAM to stop paying for AI
I bought a laptop with a lot of memory so I could run capable AI models on my own machine — private, offline-capable, and free per token. Getting them running was the easy half. The part nobody warns you about: a local model is a brain in a jar, sealed off from the internet, and giving it hands — real tools — was the hard, undiscussed work. Plus why today's frontier prices probably aren't real.
Read the story →
My dictation app wanted a subscription, so I built my own
My voice-to-text trial ran out and I didn't want another subscription for a tool I use all day. So I built the open-source alternative from source — free forever, with transcription running entirely on my Mac and most of the AI running local too. The "$40 app" that's actually free, and how I assembled it.
Read the story →
How a vague wholesaler email becomes a buy-or-pass decision — automatically
Wholesalers send deals with photos, a price, and no address — on purpose. I built a five-stage pipeline that reads the email, pulls the property's location out of the photo metadata, comps it, runs our buying math, and hands me a decision on email and Telegram. Same judgment, none of the grunt work.
Read the story →
I turned an old laptop into an always-on home for my AI agent
My assistant, Hermes, kept dying every time I closed my laptop. So I wiped Windows off an old Spectre, installed Linux, and had one AI set up the always-on server that now hosts the other. The full build — flashing the drive, the clean migration off the cloud, and keeping everything in sync.
Read the story →