~/blog/series/field-guide

Field Guide

Concrete walkthroughs for getting AI into real systems, regardless of which model you happen to be using this quarter.

What this series is about

This series is the practical counterpart to everything else on the blog. The other posts argue. These ones show.

The brief is simple. Pick a model, pick a stack, build something that works, then write down what held up and what did not. The model varies. Some posts are about Claude, some about Gemini, some about whatever happened to be the right tool for the job. The point is not the brand on the API key. The point is that working with AI in real systems is a craft, and the craft is portable.

I keep coming back to the same observation. Most of what makes an AI project succeed has nothing to do with the model. It has to do with the seams. Where does the agent get its input, what is allowed to come back out, how does the system fail when the model misbehaves. The guides in this series spend most of their time on those seams.

Read these posts when you are about to build something and want to see how someone else did it first. They are not exhaustive. They are not neutral. They are what I learned by shipping the thing, written so you can skip the parts I had to learn twice.