Field notes on building agents that earn their keep.
What I learn shipping AI agents into real operations — revenue-cycle, lead-gen, ops automation — and the open source underneath it. Written for operators, not for the timeline.
The open-source AI models worth knowing in 2026
LLMs, reasoning models, VLMs, image, voice and embeddings — license, size, and a real link to run each one yourself.
Read →The open-source stack I run an AI company on — for $0
Observability, an LLM gateway, local models, voice cloning, a $0 video path — the actual free, self-hosted open-source running the company, ports and all. Plus the honest gaps: talking-head video and social publishing.
Read →A free 24/7 server on Oracle Cloud — what actually works
Oracle's Always Free tier gives you a real server that costs nothing and never expires. We run one. The honest version: what you get after the June 2026 cut, the "Out of capacity" problem, and the retry script that gets around it.
Read →The agent is already inside your workflow
Most teams ask "where can we add AI?" That's the wrong question. The agents are already there — hiding in the repetitive judgment calls your team makes every day. Here's how I find them.
Read →