Build the systems.
Don't just use them.

Most people who "use AI" don't understand what's happening underneath. This module removes that fog. You learn how large language models process information, what tokens are, why prompts work the way they do, and what the limits of AI systems are.

The jump from "using AI in a chat window" to "calling an AI programmatically" is the most important skill leap in the program. You will write your first API call live in class. By the end you have a script that does something real.

Building agents doesn't mean writing everything from scratch. Practitioners use a stack — a set of tools that each do one thing well and plug together. You will build the same automation three different ways and learn when to use which.

Single-step agents are toys. Real agents remember context, call external services, handle errors, and recover when things go wrong. This module teaches the architecture of a proper multi-step agent.

You pick your capstone project. You scope it, design the architecture, build V1, and test it with real users outside the cohort. The practitioner acts as your technical lead. This is the hardest and most important month — where you stop learning and start building.

The cohort splits into three tracks based on capstone direction. Each track goes deep on the specific agentic patterns most relevant to their domain.

The agent works. Now make it legible to the world. This month turns a project into a portfolio piece — the kind that gets you noticed in an interview, remembered by a client, or funded by an investor.

Every student presents their capstone agent to the full cohort, all practitioners, and invited industry guests. 5-minute demo. 10-minute Q&A. All presentations recorded with student permission and published on the ISS /cohort-1 page.