AI Fundamentals
Illustrated explainers, grouped into tracks
How models read, retrieve, act, get measured, and stay safe. Each explainer covers one idea, with its own illustrations and a running example.
Designers, developers, and product managers are all product builders now, and AI is going to be part of nearly everything we build next. Knowing how it actually works is the fundamental skill, so I explain it here one illustrated idea at a time, from tokens and context windows all the way to agents and evals.
01 / 08 Why it matters
When people say AI, they usually mean a model: a program that reads the text we send it and writes some text back. The model is never the whole product, though. Somebody decides what goes into it, what happens to its reply, and what people see when it's confidently wrong, and that somebody is you.
You don't need to train your own models to build something great with them, but you do need to know what they're good at, what they cost, and how they fail.
That's why I think understanding AI is the fundamental skill for building the great ideas of the future (and for keeping them working once real people use them).
Whatever part of the product you own, understanding AI pays off:
See why AI output drifts toward generic, where taste and judgment come in, and how to design around a model's rough edges.
Learn what a model actually receives, why it costs what it does, where it fails, and how to make it reliable and safe.
Know when a feature needs an agent at all, how to measure quality before your users do, and what each answer really costs.
02 / 08 Proof of work
The stuff product builders need to know keeps changing. My first tutorials here were about Flash. Then came HTML5, JavaScript, React, algorithms, and now AI.
What hasn't changed is how I explain it, with lots of pictures, a running example, and the occasional bad joke.
03 / 08 The map
A great product needs more than a model. It also needs the web it runs on, the code around the model, and an interface people actually enjoy using, so I cover all of it. Start anywhere, because the regions connect.
Illustrated explainers, grouped into tracks
How models read, retrieve, act, get measured, and stay safe. Each explainer covers one idea, with its own illustrations and a running example.
JavaScript
Tutorials and quizzes
The language of the web, from your first variable to closures, classes, and the DOM.
React
Tutorials, quizzes, and a learning path
Components to shipping, with a learning path updated for 2026/2027.
Data Structures & Algorithms
Tutorials, labs, and quizzes
Big-O, trees, graphs, and sorts you can watch run in the labs.
Web Animation
Tutorials and quizzes
Transitions, keyframes, easing, and motion driven by JavaScript.
Canvas
Tutorials and quizzes
Drawing, animating, and handling input, one pixel at a time.
UX & Design
Tutorials
Permission prompts, dark mode, sound, and the details people feel.
Coding Exercises
Exercises and quizzes
Small projects for practice: palindromes, tip calculators, a Magic 8 Ball.
Interviews
Watch or listen
Long talks with designers, developers, and founders about how they work.
04 / 08 Start here
Not sure where to start? Pick the route that fits what you're building (or getting ready to build). Each stop is one article, and each one builds on the stop before it.
You use chatbots and want to know what's going on underneath.
You're adding AI to a product and want it to hold up once real people use it.
You want the fundamentals that coding interviews keep coming back to. Every stop has a quiz, so you can check your reasoning as you go.
Learning React instead? There's a path for that too: Your React Learning Path.
05 / 08 Quizzes
AI can recall a fact or write the code in seconds. Knowing whether that code actually fits what you're building is still your job, so my quizzes skip the trivia and ask how: which test exposes the bug, which change fixes it, and what breaks when the input changes.
What Recall. Any chatbot answers it instantly.
Which JavaScript operator means "greater than or equal to"?
>=
How Judgment. From the If, else, and switch quiz.
A speed of exactly 55. The proposed condition differs from the requirement at the boundary itself. Testing exactly 55 distinguishes an inclusive threshold from an exclusive one; large and small values alone would miss the defect.
A few things to know before you start:
Pick a subject:
Your score is just a starting point. Read the explanations, revisit what you missed, and try again. If you connect your forum account, your results save automatically.
06 / 08 Latest
Here's what I've been making lately, including new articles, newsletter issues, and long conversations with people building real products with AI:
Articles
Newsletter Interviews
From the forums
07 / 08 Community
Every product builder runs into the same walls sooner or later. Here's where we compare notes:
Thought-provoking notes at the intersection of design, development, and business, with plenty about building products with AI. It lands in your inbox every week.
Latest: KIRUPAicons is live on Product Hunt
Subscribe on kirupa.chat
Plug KIRUPA into your AI toolsConnect GitHub Copilot, Claude, OpenAI Codex, or Cursor to the tutorials and code examples here through the KIRUPA MCP server. You don't need an API key, either.
https://www.kirupa.com/mcp/
The forums
Ask questions, spot the bug, and take on a coding challenge with people who like this stuff too.
Who's online
See where people are reading from right now, live on a globe.
08 / 08 FAQ
A person! I'm Kirupa Chinnathambi, and this is my site. There's a longer answer, written in the third person for reasons, in What is a Kirupa?
Yes! Every tutorial, explainer, quiz, interview, and forum thread is free. KIRUPAicons are free for personal projects too, and a paid Pro version adds SVG sources and commercial use. The books cost money, because publishers insist.
Pick one of the paths above. If you're brand new to AI, Introduction to AI Fundamentals is the friendliest first stop.
Take a quiz! They ask how instead of what, so you'll find out quickly whether an idea actually stuck.
Subscribe to the newsletter, follow the RSS feed, or connect your AI tools to the KIRUPA MCP server so your assistant can look things up here directly.
:: Copyright KIRUPA 2026 //--