The Five Building Blocks of Every AI System

Every AI system you'll ever run into — a fully autonomous pipeline handling work with nobody watching, or the assistant helping you keep your own day organized — is built from the same small set of parts. Once you can spot them, you start seeing them everywhere, and you start knowing exactly which one to reach for when something isn't working the way you want.

In this session, we'll take one simple, relatable scenario, a customer emailing in with a problem, and watch it grow, piece by piece, from "someone reads the email and decides what to do" into a full working system. Each new piece answers a question the system runs into as it grows: how should it always behave, how should it handle this kind of work, what does it actually need access to, what should happen without anyone asking, and how do you keep the work from overwhelming one conversation. By the end, you'll have watched a real agentic system assemble itself in front of you, one problem at a time.

You'll leave with:

• A simple mental model for taking apart any AI system you run into, without needing to know its underlying framework or vendor
• How the same small set of ideas shows up whether the system is a fully autonomous pipeline or the assistant helping you stay organized
• A fast way to diagnose what's actually missing when an AI tool isn't behaving the way you want, instead of guessing at random settings
• How to start applying this today with tools you probably already have