Learning Paths

The AI Substrate — How LLMs Actually Work

The technical vocabulary to make informed decisions about agent design

Not an ML course. A translation layer between AI fundamentals and enterprise agent engineering. You'll understand what LLMs can and can't do, why they fail in the ways they do, what tokens and context windows mean for your architecture decisions, and how cost, latency, and model selection shape every design choice downstream. By the end, you'll have the technical vocabulary to make informed decisions about agent design without needing a PhD in machine learning.

Agentic Primitives — The Building Blocks of Agentic Systems

Design agents the way enterprise architects design systems: with patterns, not improvisation

Every AI agent, regardless of complexity, is assembled from the same set of primitives across four domains: the mind (how it thinks), the hands (what it can do), the voice (how it interacts), and the wiring (how it connects and coordinates). This module teaches each primitive in the order you'd actually design and build an agent, culminating in the AI Agent Canvas — a structured design tool that organizes all 17 primitives into a systematic design exercise.

A critical thread runs through this module: when not to build an agent at all. For every primitive and pattern, we address when traditional automation, a well-designed API integration, or a simple UI is the right answer instead. The most valuable skill in agent engineering is knowing when agents are the wrong solution.

Agentic Patterns

From primitives to production — the patterns that make agents work at scale

Coming soon. This module will cover the architectural patterns that emerge when you combine primitives into real-world agent systems — orchestration strategies, multi-agent coordination, error handling, governance, and the hard-won patterns that separate prototypes from production deployments.