Interactive · 16 chapters · ~40 minutes
Neural networks, transformers, memory, RAG, tool use, multi-agent systems — every idea behind modern AI agents, explained visually and animated so it actually clicks. No hand-waving, no prerequisites.
AI from the ground up — neurons, neural networks, and the transformer that changed everything.
Live 2Next-token prediction, temperature, and context windows — what's really happening when a model "thinks."
Live3The leap: autonomy, perception, reasoning, planning, action, adaptation.
Live4Perceive → think → act → observe. The ReAct pattern that powers every agent.
The superpowers — web search, code execution, files, and APIs that touch the real world.
Sensory, working, episodic, semantic — and why LLMs forget everything by default.
Retrieval-Augmented Generation: index, retrieve, generate. Giving models a library.
Pinecone, Weaviate, Qdrant, Chroma, Milvus, pgvector — where meaning lives.
Meaning as numbers, and the cosine-similarity math that finds "related."
The Model Context Protocol — a "USB port for AI" for universal tool connections.
ReAct, Chain-of-Thought, Plan & Execute, Tree of Thoughts, Reflection, LATS.
Sequential, parallel, hierarchical manager–worker, and debate patterns.
LangChain / LangGraph, LlamaIndex, AutoGen, CrewAI — and when to use each.
Prompt injection, scope creep, infinite loops, and human-in-the-loop.
Enterprise, dev, healthcare, finance, education — agents in production.
A 10-week roadmap from zero to a production-ready agent.
How to use this course
Every chapter is a visual explainer you can scroll through in a few minutes, plus a matching slide deck for presenting to a class or team. Start at Chapter 1 — each one builds on the last.