AI Agents 101

AI Agents 101 is an interactive field guide to how AI agents and large language models actually work — built for the curious skeptic. Learn from the source, not the hype. No accounts. No ads. No tracking. No internet required. Everything runs on your device. WATCH AGENTS WORK See a live two-agent Writer ↔ Reviewer loop pass work back and forth until it's approved — a feedback loop you can actually watch, step by step. UNDERSTAND THE AGENT'S TOOLKIT Tool-calling, ReAct, RAG, MCP, and multi-agent pipelines — the patterns behind every modern AI agent, each shown as a hands-on demo you can poke at. SEE INSIDE THE MODEL Tokens and the context window. Next-token prediction, one piece at a time. Attention weights across a sentence. Parameters and training with real gradient descent. Quantization. Watch the ideas behind every LLM come to life. MEANING, LIMITS & CONTROL Embeddings and cosine similarity. Vector databases. Hallucination — and how grounding fixes it. Chain-of-thought. Evaluation, benchmarks, and how leaderboards get gamed. BUILD A SECOND BRAIN Turn a folder of linked markdown notes into an AI-queryable knowledge base — that's RAG, pointed at your own notes. Explore the bundled vault of 33 linked concept notes. REAL OR MOCK — ALWAYS LABELED Every demo carries a REAL or MOCK badge. REAL means genuine logic is running on your device. MOCK means it's a scripted illustration. You always know exactly what's running — no smoke, no mirrors. RUNS ANYWHERE Works on any iPhone or iPad running iOS 17 or later. The demos use scripted mock agents, so no Apple Intelligence and no special hardware is required. Whether you're a developer, a student, or just AI-curious, AI Agents 101 replaces buzzwords with understanding — one interactive concept at a time. Free to use. If it helps you, there's an optional tip jar — but nothing is ever gated.