Ian Bigford
Back to Projects

The Gen AI Book: History to Now

3/8/2025
Repo Website

Overview

Live site: genaibook.fyi

The Gen AI Book is a free, living, interactive book about Generative AI and agents. It’s for people who want to actually understand what’s going on under the hood and ship useful systems — without the hype.

  • From “LLMs feel like magic” to “I can reason about tokens, attention, context, and agents.”
  • Bring curiosity. Leave with working mental models and simple, reproducible recipes.

Why

I built this book because I came to learn that many people haven't used an LLM yet and frankly that didn't surprise me. The space is moving crazy fast, it's fully of nerdery, doomer narratives and everything in between. So trying to do my part, I wanted to try to make it approachable from first principle to try to democratize the movement and help drop the fear around AI.

Who it’s for

  • Anyone who wants a guided journey from the history of LLMs to today.

What’s inside

  • Part 1: Foundations — what models are optimizing, how representations form
  • Part 2: Transformers — attention, tokens, scaling, why it all works so well
  • Part 3: LLM mechanics — tokenization, prompting, context windows, sampling
  • Part 4: Agents — tools, planning, memory, evaluation; MCP, Google ADK, and friends
  • Part 5: Practice — shipping, risks, and responsible use that survives contact with reality

Interactive visualizations

  • Neural nets — representation building, gradients, and loss landscapes
  • Transformers — attention maps, token flows, residual streams
  • LLM internals — tokenization, context limits, inference walkthroughs
  • Agents — tool use, orchestration patterns, failure modes

You can poke at the diagrams, watch things move, and form the “aha” that sticks.

How to read this

  • Start anywhere. Each page stands on its own, builds toward a bigger picture.
  • Short sections, minimal hand-waving, lots of visuals.
  • Links out to code, papers, and real systems so you can keep going.

Platform and stack

  • React 19 for the UI
  • Vite for dev/build
  • Framer Motion for animations
  • Tailwind CSS for styling
  • Lucide React for icons
  • Custom SVG components for the interactive pieces

Architecture and project structure

src/
├── components/                 # Interactive visualizations and UI components
│   ├── AIVisualizationGrid.jsx # Main visualization hub
│   ├── DetailedVisualizer.jsx  # In-depth concept explanations
│   ├── BookContent.jsx         # Curriculum content
│   ├── NeuralNetworkSVG.jsx    # Neural network visualizations
│   ├── TransformerSVG.jsx      # Transformer architecture diagrams
│   ├── MCPArchitectureSVG.jsx  # MCP framework explanations
│   └── ...
├── App.jsx
└── main.jsx

What you can do with it

  • Spot high-signal AI use cases in your domain
  • Design a minimal but useful agent (and iterate)
  • Ask the right questions before you commit to a roadmap
  • Communicate clearly across data, product, and engineering

Mission

Make the core ideas of GenAI and agents obvious, visual, and useful — so more people can build good systems and fewer people are scared off by jargon.

Links