On weekends I build with AI and write down what I learn.
I have a day job I like: I lead a go-to-market team. I am also a software engineer by trade, and that part of me does not switch off on Friday. So on weekends I sit down and ask two questions. What can I build with AI this weekend? And how do I apply systems thinking while I do it?
This newsletter is what came out of asking those questions for a year. The experiments, the lessons, the practical how-tos. I have built agentic solutions for sales, for marketing and for service: a search bar that sells, a chat assistant that closes, an AI receptionist with her own phone line, a factory that ships code while I sleep. Some of it works in production for real customers. Some of it taught me something and went in the drawer. I share both.
One story per issue. Each starts with a real moment: a customer types something, a friend asks for something, a number on a dashboard does not add up. Then one idea you can take to your own project the same week. Then, if you want it, the deeper version.
Some issues are short and opinionated. The long technical ones split: the front half for anyone, the back half for the people who want the wiring.
People who want to learn how to build AI solutions from first principles, and to apply systems thinking while they do it. You may be an engineer who has shipped something with an LLM and wants to build the next one better. You may be a founder or an operator who makes the architecture calls without writing the code and wants the vocabulary for what your engineers are arguing about. You may be building with an AI coding agent on nights and weekends and suspect the tool is teaching you a different way to think about software. The common thread is the question, not the job title: what is this thing for, and what system around the model actually delivers it?
There are better newsletters for prompts that will change your life. This one is about why one AI product feels like it listens and another feels like a search box with a hat on.
The model is a phone call to someone brilliant who forgets you the moment you hang up. Everything that makes a product out of that call is the system around it: what it remembers, what it is allowed to do, how you know it did the right thing, how it gets better, and where a human still has to stand. When an AI coding agent does most of the building, the same questions apply to how you work with it. That is the whole subject.
It is Sunday night, and tomorrow is Maya’s first day at a corporate office job. She types exactly that into a fashion store she likes. Result number seven is a pair of acid-wash denim shorts. The blazer she needed was in stock the whole time. Issue 1 is about what it would take for a search bar to sell, and why the answer is the person behind the counter.